195605
195605 195605
User Manual: 195605
Open the PDF directly: View PDF .
Page Count: 52
Download | |
Open PDF In Browser | View PDF |
The Position of the University in the Field of High Speed Computation and Data Handling · . . Alston S. HouseholderFree Use of the Toronto Computer, and Remote Programming of It C. Gotlieb and Others · . . c. The Mechanized Muse · .. Elizabeth W. Thomas LR.E. National Convention, March, 1956, New York Titles and Abstracts of Papers Bearing on Computers and Automation Western Joint Computer Conference, San Francisco, February, 1956 - Titles and Abstracts of Papers Now, any business, large or small, can put electronic computers to work to cut costs and increase efficiency. CD.C is proud to announce Canada's first Data Processing Centre established by private enterprise. Located at the Company's Head Office in Ottawa, this centre is available to serve government and i.ndustry in any part of Canada. USE OUR SE,RVICE .•• . If you are planning to obtain your own computing facilities, you can have CD.C test your procedures and solve specimen problems using this centre. If you have your own computing facilities, you can transfer your peak loads or overloads to CD.C If you cannot justify having your own computing facilities,... you can make use of those at this centre. If you are confronted problem, you may find nor sufficient facilities Computing Centre will with an unusually difficult research you have neither the ne'cessary staff in such a situation, the CD.C be found invaluable. If you have continually recurring computing or data prOcesssing p·roblems, you can arrange to leose communication facilities linking your office directly wnh the CD.C Centre. Highly qualified and experienced applied mathematicians and engineers, backed by the latest computer facilities, both digital and analog, are at your service for problem formulation and analysis, programming, computation, data reduction, application of data processing equipment to business procedures, practical instruction in computer operation and 5604R COMPUTERS CYBERNETICS Vol. 5, AUTOMATION AND ROBOTS • AUT,O'MATIC CONTROL No.5 May, 1956 ESTABLISHED SEPTEMBER, 1951 ARTICLES ANn PAPEHS The Position of the University in the Field of High Speed Computation and Data Handling Free Use of the Toronto Computer, and Remote Programming of It,Part 1 ••• A. S. Householder 6 ••• C. C.. Gotlieb and Others 20 ...... E. W. Thomas 11 FICTION The Mechanized Muse REF~RENCE INFORMATION I.R.E. National Convention, March, 1956, New York -- Titles and Abstracts of Papers Bearing on Computers and Automation Western Joint Computer Conference, San Francisco t February t 195.6 -Titles and Abstracts of Papers New Patents 14 26 ••• R. R.. Skolnick The Editor's Notes Index of Notices Advertising Index 40 4 4 50 Advisory Committee: Samuel B. Williams, Herbert F. Mitchell, Jr:, Justin Oppenheim Editor: Edmund C. Berkeley Assistant Editors: Neil D. Macdonald, F.L. Walker Con tribu ting Edi tors: Andrew D. Booth, John M. Breen, John W. Carr, III, Alston S. Householder Publisher: Berkeley Enterprises, Inc. 513 Avenue of the Americas, New York 21, N.Y. - Algonquin 5-7177 Advertising Office: 36 West 11 Street., New York, II, N.Y. - Gramercy 7-1157 Editorial Office: Main Office: 815 'Washington Street., Newtonville 60, Mass. - Decatur 2-5453 or 2-3928 Advertising Representatives: Sa~ Francisco - W. A. Babcock, 605 Market St., San Francisco 5, Calif. Yukon 2-3954 Los Angeles - Wentworth F. Green, 439 So. West'ern Ave., Los Angeles 5, Cal if.' Dunkirk 7- 8135 elsewhere - the Publisher CDMPllTERS AND AUTOMATION is published monthly. Copyright, 1955 by Be.rkeley Enterprises, Inc. Subscription rates $5.50 for p.ne year, SlO.50 for two years, in the United States; $6.00 for one year, $11.50 for two years, in Canada; $6.50 for one year, $12.50 for two years elsewhere. Bulk subscription rates: see page 10. Advertising rates: see page 48. Entered as second class matter at the Post Office, New York, N. Y. - 3 - THE EDITOR'S NOTES THE PRINTING OF TECHNICAL INFORMATION IN "COMPUTERS AND AUTOMATION" I. From Mr. I. McNamee and E. Fullemfider, u.s. Naval Ordnance Laboratory Corona, Calif. Thank you for your letter of Feb. 6, and your interest in our subject of filtering s~m pled functions. We would be honored to have our paper appear in "Computers and Automation". Your suggestions or questions, or those of your readers, would be valued by us, since we are still actively interested in the general problem. Thank you for your kind invitation to publish our paper. II. From the Editor to Mr. McNamee and Mr. Fullenwider Thank you for your letter of March 26, enclosing your paper "Filtering Sampled Functions". We think your paper is interesting, and would like to publish it, if you could perhaps take on the labor of typing it in the fo~ in which it could be photographed for photoof set reproduction. We make this request because 'your paper at present covers 41 pages of doubled spaced manuscript, ~tt by 11", including 4 full pages of figures, 38 pages bearing mathematical symbols, and several tables. The procedure for preparing master copy for photooffset for our purposes is quite simple. There is only one main rule: type it exactly. right with single line spacing in columns four inches wide. The rest of the procedure is given below. We ask you to do this particularly b ecause we h.ave only one person at present in our group who can intelligently copy you r mathematical symbols, and who has satisfactory handlvriting and that person is quite busy with some other tasks. III. Comments from the Editor As readers of "Computers and Automation" will have noticed, tve have taken full advantage from time to time of the fact that we publish our magazine by photooffset. Quite often for example, we have been able to print quickly the titles and abstracts of a computer conference or meeting because lve have received .- 4 - three copies of the printed program. We could then cut up two copies to make a one-side set to be photographed, keeping the third'copy fur reference and checking. The narrow columns of most of the programs of computer meetings are ideal for making columns in our pages. We are eager to make our publishing' fa~ cllities\available to authors who have papers relating to computers and their applications and implications. If anyone desires to have his paper published somewhere else, we are completely satisfied. But when there is no room somewhere else for his paper, and if it appears suitable for the rather "un-fencedin" editorial coverage of "Computers and Automation", we shall be glad to consider his paper, in preliminary form or final form, for publication. It will be very easy for us to say yes to publishing a technical paper if it is sent in to us in good form for photooffset reproduction. IV. Procedure for Preparing Master Copy for Photooffset For technical information, to be published in our pages in two columns, the reduction in size that we regularly use is 2~~. In o~r words, an actual measurement of 10 inches on the actual copy is reduced photographically to 8 inches in the film from which the magazine is printed and other distances proportionately. {continued o~ page 381 • * -'f INDEX' OF NOTICES For Information on: Advertising Index Advertising Rates and Specifications Back Copies Bulk Subscription Rates Computer Directory Manuscripts Reader's Inquiry Form Special Issues See Page: 50 48 46 10 44 31 ,50 3Q Address Changes: If your address changes,please send us both your nelf and your old address, (torn off from the wrapper if possible), and allow three weeks for the change • Always tense but never tired Again CTC comes up with an advancement for more secure, more effective eleetronic assemblies. It's the new Perma-Torq* constant tensioning device for tuning cores of standard CTC ceramic coil forms. CTC's Perma-Torq, a compression spring of heat treated beryllium copper, has very high resistance to fatigue and keeps coils tuned as set, under extreme shock and vibration. It allows for immediate readjustment without removal or loosening of any mounting nut or locking spring. But most important of all - Perma-Torq like all CTC components is quality controlled. CTC's quality-control means you get consistent top quality components. Each step of production is checked, each component part - even though already certified - is checked again. And finally CTC's finished product is checked. That's why CTC can offer you a guaranteed electronic component standard or custom - whose perform·· ance you can depend upon. CTC researchers and practical experts are always available to help solve your components problems. F{)r samples, specifications and prices write to Sales Engineering Dept., Cambridge Thermionic Corporation, 437 Concord Ave., Cambridge 38, Mass. On the West Coast contact E. V. Roberts, 5068 West Washington Blvd., Los Angeles 16 or 988 Market St., San Francisco, Cal. NEW PERMA-TORQ UNITS come completely factory assembled to mounting studs, elimmating the bother of assembling and adjusting separate locking springs. CTC coil forms with Perma-Torq Tensioning Device are designated PLST, PLS5, PLS6 and PLS7, are completely interchangeable with the LST, LS5, -LS6 and LS7 series, and are availa.ble at no increase in price. ·Pgtent pending ...'"'~ ~ :~~' \"'Y''''~'',«,} *, .....M~'#' CAMBRIDGE ,.,. ~,' "'"wA"»"Jo.'...... .:.' '",.-., " ..",.; THERMIONIC CORPORATION makers 0/ guaranteed electronic components custom or standard THE POSITION OF THE UNIVERSITY IN THE FIELD OF HIGH SPEED COMPUTATION AND DATA HANDLING ALSTON S. HOrSPHOLJaFR ~athe~atics Panel Oak P.idge National Lahoratory Oak ~id~e, Tenn. It seems necess ary to begin this t a l k with what may seem a platitude by saying that the position of the university, in this field as in any other, can be stated very s imply. It is a source of instruction and a center of research. If a subject is already closed, or is so trivial and unimportant as to be u ndeserving of continued study and deve lop men t, then it has no place in the university c u rriculum. And a presentation that goes nofarther than the formal development. of the known, that opens no pathways to the unexplored, such a presentation may be appropriate to the trade school or the catechism, but not to the u n iversity. To quote A. N. Whitehead: "So far as the mere imparting of information is c 0 ncerned, no university has had any justification for existence since the popularization of printing in the fifteenth century". And elsewhere in the same essay: "Fools act on imagination without kn01dedge; pedants act on know ledg e without imagination. The task of a university is to weld together imagination and experience". It seems logical to develop the s ubj e ct before us by first sketching the position that has been occupied and is occupied by uni ve rs i ties and other ins ti tutions 0 f h i 9 h e r learning; and then passing to the future with some predictions and inj unctions. The par t that has been played by universities in thi s country in the development of digital computers is well known; it will be enough to mention a few high spots. Unfortunately, in the early stages the work was veiled in wartime secrecy, but fortunately the veil is now long sin c e removed. Outstanding early contributionsw6re from the University of Pennsylvania 0 nth e Eniac, the Massachusetts Institute of Technology on the Whirlwind, and Harvard University c n the several Marks. Long before this, the logical possibility of constructing a gener a 1 purpose computing machine had been demonstrated by A. M. Turing in his doctoral disser·t a ti 0 n l'lritten at Princeton. Turing's pape r wa s published in the late 30's, and in the earl y 4U's McCulloch and Pitts, at the Universities of Illinois and of Chicago, published a paper which carried Turing's ideas somewhat further by shotdng that one could go a long 1'1 a yin interpreting the vertebrate central n e rvo u s system as a Turing machine. There have be en subsequent developments in this direction a t Chicago, at MIT, and at Southern Californ i a in particular, and while this work is off the - 6 - ma i n line of practical computer developm en t and application, it could turn out to have an influence on the machines of the future. To return to the main line, the Institute for Advanced Study and the University of Illinois are next to be mentioned. Probably no one person has been more influential on the log i cal design of current machines than von Neumann of the Institute. It is unfortunate that the early reports by von Neumann, in collaboration tV i t h Goldstine, Burks, and others, on logical design and on programming, did not appear in the open Ii terature, but they had a wide cir cui at ion nevertheless. The classical paper by von Neumann and Goldstine on the analysis of erro r s in digi tal computation did appear in the Bulletin of the American Mathematical Society, and i t laid the foundation for the study of this hitherto undeveloped, but now critically important field. A number of machines, including two built at the University of Illinois, have followed more or less closely the design laid down at the Institute for Advanced Study. Somet'lhat less ambitious development and construction projects exist or have existed at other universities: California, Pennsylvan i a State, Michigan, Wisconsin, and perhaps others. In the field of applications, the University of California at Los Angeles was able to p r 0vide housing for the Institute for Numeric a 1 Analys is. More recently it took over the operation of the SWAC (Standards' Western Awwmatic Computer), and continued to maintain asanewhat reduced group tvhen INA was discontinued by the National Bureau of Standards. Consequently, this University has for some time been an active center of research in numerical analysis. To a lesser extent the proximity of computi n g machines has stimulated work in numerical analysis in places like American University, t h e University of Maryland, the University of Delaware, to mention only a few. Finally, as you all know, commercially made machines are coming increas ingly 1\'1 thin reach of even the academic budget, and so many schools are taking advantage of the fact that I could not hope to n a me them all and hence refrain from mentioning any. With this very sketchy summaryofuniversity activities in this country, let us look abroad where the situation may be less familiar. First, as you may recall, the Williams tube was develoRed at the Uni vers i ty of Manches ter, and the Computers Rnd Automation effort has been made to devise a list of operations that make programming as simple as possible, since the machine is to be made a va i 1able to outs iders lvho tvill do their 0 If n p r 0gramming. The name of Stiefel, at Zurich, is perhaps best known in this country when coupled with that of Hestenes, as codiscoverer of a method of successive approximation to the s 0lution of a system of linear equations, wit h convergence after a f ini te number of s t e p s • More recently he has made important contrib utions to the theory of successive approximation in general for linear systems, or, as he calls them, methods of relaxation. The name of R utishauser is perhaps less well known, but a series of papers on what he cal.ls the quotientdifference algorithm has attracted considerable attention. In these papers he succeeds in tying together and generalizing in a remarkable way a multitude of seemingly disparate tec hniques in numerical analysis, such ascontinued fractions, the Bernoull i met~od of sol v i n g equations, and the Aitken 5 - process for accelerating convergence. In this general connection, mention might be made also of s 0 m e papers by F. L. Bauer, at the Technische Hochschule, Munich, who has shown how the metho d of Bernoulli, which converges linearly, can be modified to yield quadratic con verge nc e such as is to be had by the use of Graeffe's method. Thus it is possible to achieve the rapid convergence of the one method wi tho u t sacrificing the self-correcting feature of the other. prototype of the Ferranti computer was c 0 nstructed, of which one of the early models is located on this continent at the University of Toronto. A second machine for computing use has been completed at Manchester and also, as a developmental project, a transistor machine. Cambridge University has been outstanding in both development and application. The Edsac I was one of the earliest of the h ig h s pee d machines of t~hat might be called contemporary design, to distinguish it from the more primitive Eniac. It is of interest, too, t hat Leo, the machine belonging to the J. Lyons and Company, and used for accounting, Ivas built at Cambridge, largely following the design of Edsac 1. The Wilkes-Wheeler-Gill text on programs was the first on this subj ect to rea c h the commercial press. Also, the Edsac grou p has been otherwise influential in that several of its members have been in this country 0 n extended visits to MIT, Illinois, and e Is ewhere, and Cambridge is a Mecca for Americ an experts going abroad. Ed~ac II is one of the . first to use magnetic cores throughout and the design of the control makes use of the scheme of microprogramming devised by Wilkes. The same scheme, in a form revised somel~hat by Billing at the Max-Planck Institute, is being follO\~ed in the core machine under construction at Gottingen. I do not hope to be exhaustive and hop e not to be exhausting. Hence, I pass over other Bri tish acti vi ties and proceed to the continent. Starting at the far North, a relay machine, the BARK, was constructed at the Technica 1 High School in Stockholm by the Stvedish Board for Computing Machinery. This \Vas followed by the BESK, with cathode ray tube storage, and upon completion of the BESK, the BARK tfas dismantled. The Besk remains the fastest machine in western Europe, and so far as I know, Ifill be surpassed only by Edsac II in the n e a r future. The machine is in the hands of a small but able group of mathematicians and engineers. Tl)ere are plans for making copies in Cop e nhagen and in Oslo. Interest in digital machines was rathe r s low to develop in Germany, but nOt~ it is extensive. Last October there was held at the Technische Hochschule, Darmstadt, a three-day conference on digital computers and data processing. About 600 people were in attendance, and al though many different countries tV e r e represented, probably 80 percent or more 0 f. the participants were from Germany, and 0 f these, most were from uni vers i ties and the technical high schools. At the Max-Planck Institute in Gottingen a small tape-controlled ma~hine, the G 1, has been operating for some time, and three other machines are currently under construction. One is the core machine already mentioned. Another is an enlarged version of the G 1, and the third is a more standard drum type. Drum machines are also being built in Munich and i.n Dresden, at the Technische Hochschule, and at Darmstadt a small core memory will be backe d up by a .drum. In the way of applications, the Munich group seems to be rather more in t e rested in bus iness than is ·the case elsewhere. It is interesting to note that in Germany the Williams tube is being bypassed compl e t e Iy , all storage being magnetic, with either dru m or cores. At ~terdam, the Mathematical Center of the University constructed a relay machine,the ARRA, and then proceeded by additions and replacements to turn it into an electronic m achine. They are also working on a mach i n e wi th magnetic core memory. The Mathematic a 1 Center and the Department of Mathematics seem to be better coordinated than is the cas e in some schools, and Amsterdam can be classed as one of the leading centers of research in numerical analysis. Farther South, in Zurich, a relay machine has been in operation for some time at the Technische IIochschule, and a drum machine i s being buil t. In designing the latter a special - 7 - Computers and Automation While the Darmstadt Conference was interesting enough in providing a vielv of d eveloprnents in western Europe, the occasion lvas made more dramatic by the presence of several p a rticipants from the other side of the Iron Curtain. I have already mentioned Dresden in East Germany. From Prague came Svoboda and Oblonsky, who have built a drum machine, the SAPO, and are building a more elaborate one. The SAPO~, in effect. three mutually checking mach i n es, and it utilizes five-address commands, the extra two addresses designating alternate loc ations for the next command. Svoboda's g r 0 u p seems to be very active in numerical analysis, and in logical synthesis, but unfortunately, although I have some of their publications in my possession, they are in Czech so t hat my understanding is something less than complete! From Moscow came Professors Lebedev and Basilevsky, accompanied by Drs. Ktorov and Novikov, l-vho acted mainly as interpreters. To the best of my knowledge, this was the fir s t time the West had been informed of Ru s s ian activities. It turns out that they h av e a drum machine, the Ural, and a machine calle d the BESM which uses cathode ray tubes and auxiliary drum and tapes, with triple address commands a'nd floating-point binary representa,tion of numbers ~ It operates at a speed that surpasses anything else exc~pt the NORC. As I remarked before, I am not trying to be exhaustive, and perhaps I should brin«this account to a close. It is intended 0 n 1 y t 0 suggest the extent of the activi ty on the part of educational institutions, and to indic ate some" of the things they are doing. In comparing the scenes in Europe and America, however, there are several important factors that should be borne in mind. In Europe the educ ati 0 n al institutions are all financed by either the national or the local governments. Moreover, although England has, for example, a National Physical Laboratory, which corresponds roughly to our National Bureau of Standards, tot h e best of my knmdedge there is no c ompar abl e ins ti tution in the countries of conti n en tal Europe. To some extent, its place is t a ken by the schools. It 'is time now to come back home and cast a glance into the future. In this country no school needs any longer to build a machine just to have one, and countless schools are finding it possible to acquire them by purchase, b y rental, or even as outright gifts. In so m e cases construction may be worthwhile as a research or a training project. But it seems to me that in the future the universities can contribute most in applications and in training. In the matter of training, one he a rs 0 n all s ides of the shortage in 'technical manlXJ.fer, and in the field of computing the shortage is, if anything, even more acute than elsewhere. Until quite recently there has been no demand to speak of for people in this field, whereas, now the demand is very great and several conferences have been held for discussing t II e specific problems associated tv! th training of computer people. Personally, I am not sur e that it is proper to approach the problem i n just this tvay. The shortage of techni call y trained people is general. Th at there i s a shortage of computer people is but a corollary to the main theorem. HOtfeVer important the special case may be, it remains but a special case. The universities cannot be blamed for 'the shortage and there is relatively 1 itt 1 e they can do to alleviate the situation in addition to what they are doing now. Certainly, they should not resort to narrOtV spec i a 1 ization. In this country much of the early stimulus to commercial development came from government laboratories, and some came from i ns u ran c e firms and other business organizations. In this respect the experience in England has bee n comparable, but on the continent there h as been nothing of the sort. Consequently, a few British firms are making computers, but there have been none on the continent. IIence, s pe a king generally, Amsterdam, Zurich, Darmstadt, and the others had no choice but to build 0 r go without. This situation is changing, but the changes are very recent. -, 8 - Let me be a little more specific. In our organiz ation at Oak Ridge, tye operate a digital computer called the Oracle. We, therefo r e , have a staff of analysts, programmers, and coders. Al though the distinction bet11een the jobs of analyst, programmer, and coder are fairly clearcut, all our programmers do their own coding, and often our analys ts do likewise t Our general policy in hiring is, therefore, to require at least a bachelor's degree with a mathematics major. Naturally at the start it t1TaS seldom possible to find people with any machine experience or any special tr a i n in g along these lines. It is still not easy. But on the other hand, I Was never much concerned by the fact. Basic ability and a good gener~ mathematical background,seem to me to be much more important than special experience. since the latter can be acquired on the job by anyone equipped with the former. Therefore, if the record of a new graduate shows a co u r s e in programming, or even in numerical analysis, I would not be greatly impressed by that fact alone. This is the negative side of the picture. But there is also a positive side. I do think digi tal computers are here to stay and t hat the fact should have a real bearing 0 nth e college courses, especially in mathema tics. Computers and Automation The bearing on courses in physics or che~try or other fields from which problems a r i s e Th e would seem to be at least less direct. advent of the digital computer is 1 ike 1 y to stimulate developments in certain are a s i n physics, say, and this would naturally be reflected in course content, especially in the more advanced courses. But whatever the source of a computing problem, the analysis and the programmdng is basically mathematical in character. the textbooks. Newton's method is generally presented as though it applied to real roo t s only, whereas in fact it is quite general. And there is much interesting material availableon the geometry of the roots in the complex plane, all of which is good mathematics and potentially useful to anyone who might ever have occas ion to solve an equation numerically, and this includes physicists and chemists as well as n umerical analysts. My suggestion is that the addition of one or two or three special courses in programmdng or numerical analysis or both lsnotsufficient and may even be detrimental, and that instead there should be a rather general reorientation of all the courses. This go s pel has bee n preached on other occasions, and I may be boring those who have heard it before, but nevert h eless, a few specific suggestions migh t be a r repeating. Perhaps some are already b e i n g followed in some places, but not all of the m are in effect everywhere. The traditional first course in differential equations devotes a great deal of time to the search for transformations and integrating factors that reduce the solution to a quadrature. I conj ecture the theorem that in the class 0 f equations which arise in practice the technique is almost everywhere inapplicable. Moreover, even if one does succeed in reducing the problem to a quadrature, the quadrature itself can seldom be carried out, or if it can will g e nerally lead to unmanageable functions, so that one is ultimately forced back to a direct attack upon the equation itself. In the case of first-order linear equations, or of a few equations reducible to that form, the technique is useful, but I can think of no other outstandkg examples. In place of a study of integrati n g factors I would suggest two things. The first is a little time spent applying an elementary numerical method, say that of Euler, for a rough And numerical solution of a few equations. the second is an introduction to the Poincar~ Bendixson theory of singular points. E i th e r one, or, better yet, the two together, can, it seems to me, give the student a much bet te r feeling than integrating factors can 0 f the relation of a family of functions to the d i fferential ~quation it satisfies. ' I have less of a quarrel with the theory of equations, but even there an instructor may leave the student tvi th the impress ion t hat Cardan's formula has some direct practic a 1 utility. But there is, for example, an interesting and important theory associated wit h Bernoulli's method of solving an equation, especially as developed in the recent papers by Bauer and Rutishauser already mentioned, and in papers by Aitken and others, and yet the method itself is seldom or never developed in In calculus it is easy to leave the student with the impression that any integral, or a t least most integrals, can be carried out analytically by someone smart enough to do so, and that one is always better off when the attempt is successful. The fact is, of course, that many functions are best evaluated by carryi n g out a numerical quadrature, and if a functio n is given initially as an integral the likelihood is strong that this is such a function. In the treatment of limits the attention is generally focussed upon the conditions that assure the existence or nonexistence of a limit. In practical work, in cases where a limit is known to exist, one is greatly interested in estimates of the deviation from the limit of an arbi trary term in the sequence. These estimates are given by remainder formulas. They are, indeed, often useful in proving convergence and their consideration requires little more than ash if t 0 f emphasis. Finally, I wish to enter a plea fo r the earlier introduction and more general use 0 f matrices and vectors. It seems to me that this introduction comes most naturally in fhe freshman course in analytical geometry. The i r importance, for both pure and applied mathematicians, seems scarcely open to question, and yet a deplorable cultural lag permits even s 0 m e mathematics majors to graduate with the barest nodding acquaintance. Digital computers and computing are drawing upon a variety of areas which are usually taken up only in the more advanced courses i n mathematics, but which could be, and sometimes are being, at least introduced at more elementary levels. Foremost among these are Boolea n algebra and symbolic logic. Another i s the theory of groups. Still another may be co~ atorial topology, since apparently Gmriel Kron has had spectacular success in applying this in solving certain systems of equations. Whether such topics should be offered in special undergraduate courses, or whether the entire mathematical curriculum should be reorganized, as is being done in some places, is a question Ishall not attempt to answer here. A while back I indicated that I did not consider an undergraduate course in numerical analysis too important, but now, as you see, I am virtually advocating that the entire mathe- - 9 - Computers and Automation of application, both in their training and in matical curriculum be oriented toward numerical analysis. My argument is that the slant can their research, although their training should increase the utility of the subj ect for a l l never descend to the trade school level. While some background in physics, chemistry, andofuer those who are interested in mathematics 0 n 1 y as a tool, without in any tfay lowering the digsciences is certainly advantageous to the future programmer or numerical analyst, it seems nity of the courses as mathematics. Where such to me that eventually his problems become math. a viewpoint does not prevail, however, a course ematical in character. Hence, the immediately in numerical analysis can be of value provided it is not a mere collection of recipes. relevant training and research should cente r in the mathematics departments. This plac e s upon them a unique responsibility not sh are d To illus trate t"hat I have in mind her e, suppose in a course in numerical analysis 0 n e by other departments. is about to take up the solution of algebra i c equations by the method of Bernoulli. In this In all this I have spoken only of scienmethod one forms a basic sequence which satistific uses of the machine and have skillfully fies the linear difference equation whose coavoided mentioning bus iness applications. This efficients are those of the algebraic equation is for the very good reason that I h a v e no to be solved. From this basic sequence 0 n e background for discussing this area. Neverforms a secondary sequence by taking ratios of theless, on general principles it seems th a t consecutive terms. If the equation has a single the problems are still logical and arithmetical root of largest modulus, then the new sequence in character and that mathematics could help. approaches this root as its limit, and the rate The hard thing is to persuade mathematicia n s of t.)nvergence is determined by the rati 0 0 f to become interested in business problems, and the modulus of this root to that of the n ext business experts to study mathematics. I prelargest root. If there are two largest root s dict most rapid progress tvhever mathematicians, of equal modulus, this secondary sequence has engineers, and business experts can be p e rno limit, but other sequences can be for m e d suaded to join together in arriving at a common whose limits are the coefficients of the quadunderstanding by which to differentiate the ratic satisfied by these two roots, and the primary needs of business from the incidental rate of convergence depends upon the ratio 0 f byproducts of established procedur e s, and thence to devise the hardware and routines for the common modulus of the two roots to that of achieving the real objectives. Where coul d the next. If the two largest roo t s a re of such teams form more readily than in a univernearly equal modulus, convergence to the 1 a rsity? gest will be slow, but convergence to th e coefficients of the quadratic may be sui tab 1 y -_ FNl ::= _ _ _ _ _ _ _ _ _ ... _-_ _ _ _ _ _ _ _ _ =:c fast. Now all these points can be brought out and illus~ated by numerical experimen t a tio n with a fetf simple, manufactured cub i c san d quartics with no very laborious and extensi v e BULK SUBSCRIPTION RATES computation. In short, I am recommending that the course be one in numerical analysis, d i rThese rates apply to SUbscriptions coming in ected toward the mathematics, rather than 0 n e together direct to the publisher. For example, in numerical techniques for developing special if 5 subscriptions come in together, the savskills. ing on each one-year subscription tv'i 11 be 24 percent, and on each tt-lTo-year subscription will Perhaps some apology is due for a t 0 0be 31 percent. The bulk subscription rat e s, frequent use of the first person singular i n depending on the number of simultaneous s u bthis account. The excuse is that the viewpoint scriptions received, follow: is ~ own personal one, and it is very likel y to be limited and onesided. I have objec ted Bulk SubscriPtion Rates to specialization and yet I have proposed a (United States) fairly general slanting of the rna th emat i c s courses toward numerical analysis. In feeling Rate for Each Subscription, and Number of Simul taneous Resulting Saving to Subscriber that such a slant would be quite gener a l l y Subscriptions One Year Two Years beneficial I may be merely exhibiting the bias of my profession. But to summarize: Universi$6.60, 37% 10 or more $3.80, 31% ties, both here and abroad, have played an im4.20, 24 7.25, 31 5 to 9 portant part in the development of dig ita 1 4.60, 16 8.00, 24 4 computing machinery, and they have contributed 5.00, 8.80, 16 3 9 to the understanding of how to use them. Cer2 5.25, 5 9.55, 9 tainly, government laboratories and commercial firms have also contributed, but my topic conFor Canada, add 50 cents for each year; 0 u tcerns only the universities. In th e f u t u re, side of the United States and Canada, add $1.00 in this country at least, it seems to me that for each year. universities can contribute mo~t in the field - 10 - TIlE MECHANIZED MUSE ELIZABETH W. THOMAS 'led Bank, N.J. Dr. Ca'rl Jonus Yaffee, Ph.D., M.A., M.S., Associate Director of Cornumbia University's School of Engineering and creator of the r ecently unveiled Yaffee Electronic Relay Poe mWriter, is a fresh-faced, blue-eyed silvery haired gentleman of 67, wi th a boyish s mil e and a diffident manner. He looks more 1 ike a busy and contented small-town pediatrician than an internationally celebrated scie n tis t with nearly half the letters of the alphabet following his name;' but YERP, as his current brainchild is more generally known, is the sixth in a series of notable contributions to the ar t s and sciences which he has made since his arrival in this country from Austria in 1918 - contributions which include such familiar househo 1 d devices as the Yaffee Automatic Pinochle Player (YAPP), the Yaffee Automatic Check-Book C 0 rrector (YACC)', and the Yaffee Electronic Automatic Housewife (YEAH), the latest mod e 1 0 f which not only orders and cooks the meals but washes the dishes and puts them away. YERP has been five years a-borning, and has employed a full-time staff of 55, composed of members of Cornumbia' Engineering, Applied Science, Mathematical, and English Departmen~ -- technicians, and clerical help. It is, says the Doctor modestly, merely an extension of the gigantic electronic "brains" wh!ch are be in g produced regularly by our industrial an d university'engineering laboratories. YERP embodies many of the"same principles and ,muc h of the same circuitry as its mathematical relatives; it differs mainly in that it treats w 0 r d s rather than mathematical data. It can be demonstrated, says the Doctor -- adding, wit h a twinkle, that he viII not do 'so just now -that the mechanical processes involved in the cons truc tion of a poem are essen t i all y th e same as those associated with the solution of a series of complex equations. The prob Ie m was chiefly one of providing greatly expanded maane to-electronic storage capac! ty, t 0 accommodate YERP'S enormous "vocabulary"; i ncreased sensi tivi ty in selectors, by which YERP is enabled to "recognize", and select or reject a work; stepped-up operational speed, permitting YERP to "scan" almost infinite combin ations of words in a few fractions of a second; and a simple but flexible coding system, whereby YERP'can be "instructed" to move along p r eselected operational procedures. At the present moment YERP occupies three rooms of the Cornumbia Physics Laborator y in the Morton Memorial Building, but i t IN i l l shortly be installed in a specially built annex of the New York General Library, where it will be available to poets and students a t a nominal fee. Pending its removal to permanent quarters, visitors may be taken on a tour of YERP any Thursday afternoon, upon application to the University· secretary. On the occasion of my audience with YERP I was fortunate enough to have the good Doctor himself as interpreter. My group contained, I suppose, the average mixture of literary, scientific, and miscellaneous components: representatives of the Book Departments of t h r ee Metropolitan dailies, two engineering students from ne ighbor ing uni vers i ties, a young re porter from a well-known weekly news magazine, a lady poet from Ohio, two gentleman poets both from New Jersey', and myself. We were ushered into a comfortably furnished reception room in the Phys ics Laboratory, where the great man greeted us informally and quickly disarmed the fears of the mechanically unsophisticated. "The scientist uses beeg words to impress you wi th his knowledge," he began. "He r e at Cornumbia ve do not strive to impress -- v e seek to instruct. Anybody among you who e e s able to chenge a fuse will have no dif~lc-ul ty in following my brief remarks on the IOOchanlcal side of our machine', which viII be of interest, I hope, alike to poet and engineer. The latter, if he desires, may satisfy his scientific curiosity by reading· the technical manual 0 n the machine, which ees available at t his department; and the former may wish to purchase the little book of poems produced by the machine, which ees on sale at the Un i v e r sit Y Book Store." We learned that YERP contains over 50 0 vacuum tubes, 10,000 transistors, 20,OOOcrystal diodes, 350 pluggable units, 1,356 mile s of wiring, a magneto-electronic storage un i t with a 5,OOO,OOO-word capacity, a prin tin g unit capable of printing 1000 words a minute, and a vocabulary of about 1,500,000 1'J 0 r d s which are punched on cards. The vocabu 1 ary, incidentally, is constantly being revised and kept up to date by a special group of readers - 11 - Computers and Automation ttl think it's sweet, n Ivhispered his compatriot. We gathered around the table and watched carefully. Willy looked perplexed. "HOlY abo u t a sonnet?" suggested one of the newspapermen. "B - U - S - T - E - R , " spelled out the Doctor, slowly and clearly. ttSonnet all right with you?" ask e d the Doctor. "B - U - S - T - E - R, " repeated Mis s Matthews, 'a click of the instrument's key s punctuating each letter. "OK 11ith me," said Willy. don't it?" "It rhyme s, "Proper noun,t1 said the Doctor. "Iambic pentameter?" asked the Doc tor. "Or do you prefer one of the more mod e r n modes?" Click, went Miss Matthews. "Rhyme-code-punch 3 in column 27," c tinued the Doctor. "Iambic pentameter," said Willy firmly. "I'm a fool for Iambic pentameter." 0 n- Click. The Doctor flipped fourteen switches 0 n the machine's control panel. Each switch, he told us, called for a line in Iambic pentameter. "Ve viII now select the rhyme-pattern," said the Doctor, t1and impulse the cor r e c t slvi tches. Thus ees the machine en a b 1 e d to select terminal v.ords in the desired rhymin g sequence." "Route to Selector Number three," sa i d the Doctor. "That's the obligatory selector," he explained to us. "The given '\ford must then occur at least once in the verse." Click. "Re-punch and compare," concluded the Doctor. Mter a certain amount of prompting, Willy selected a standard Shakespearean rhymiQJ pattern , and the Doctor adjusted fourteen more switches to the corresponding positions. Clickity clickity clickity sssswich plunk, went the word actuator, and Miss Matthews held two cards up to the light and compared them. tlOK," said the Doctor, "ve're ready to -what you say, roll." "Now," he sa ide "If there are any special vords, such as vould not be included i n the standard voc abul ary, ve may now ins e r t them. tt He inserted the card in the input u nit, depressed the "ON" switch, and the card loY a s promptly swallowed up. We were then conducted into a third room, where the printing mechan-. ism was housed. Willy regarded his shoes bashfully. "Think of this purely as a s cie nt i f i c demonstration," said the Doctor encouragingly. "The young lady's name, perhaps?" "Name of Beatrice," said Willy, blushing furiol!~~y. "Now," said Dr. Yaffee, "before ve tu r n on the sldtch, one vord of caution. Ve mus t not ask too much of our machine. Just as the diamond comes rough from the mine, a If a i tin g the craftsman's hand to give it 1 us t e r an d beauty, so comes the poem from the machine. tI "H'm," said the Doctor dubiously. "I always call her 'Buster''', volunteered' Willy, and the Doctor brightened. Willy was instructed how to turn the machine on, which he did. There was a cliCking and '\fhirringi lights on various panels around the room flashed on and off, and we waited for perhaps three minutes in silence. Th en the type bars started to move. ''Miss Matthews," he called. Apr e tty, dark-haired girl materialized from someWhere, and sat down at a table holding an instrument that looked like a small typewriter. She took a blank card from a rack, similar to the ones Ive had seen in the drawers, inserted it in the machine, and awaited further instructions. we read. "Please to observe closely while the young lady prepares the card for insertion in the machine," said the Doctor. "The instrume n t she ees using ees called the Word Actuator." "Stop, stop," bellowed the Doctor, leaping for the control button. There was instant cessation of the hum of machinery as the type bars subsided in their bed. "Elizabeth! Car- AS ANTIQUE ART ATTIRES AN AURA'S ARC AND AIDS AMBIGUOUS ALE AS AN ADJUSTER - 12 - Computers and Automation called "Prospectors", whose job it is to read tion and classification of our vocabulary; he all of the poetic output of the count ryan d worked for five years to develop the prese n t report on all significant trends in word usage. system, and ve may confidently say that f e Ttl At present, for instance, it is practically known human emotions are not included in it. obligatory for the poet to employ such 1.ford s So pick your sobject, and ve viII procee d as "nubile", and "incandescent", at leas t OIre without further delay to write our poem." in each poem. Willy appeared to be overcome by s t ageTime required for the produc t ion of a frigh t. "I thought you could jus t pus h a standard-length poem is approximately f i v e button, " he quavered. minutes, exclusive, of course, of the tim e consumed in preparation and polishing, wh i c h "Come, come," rallied the Doctor kindly. may take from ten minutes to two hours, d e"A young man like you, at the peak of his sexual pending upon the individual operator. maturi ty -- sure ly there mus t be some 0 net 0 1.'Vhom you 1.'Iould 1.·dsh to express yourself -" "It ees not true," said Dr. Yaffee at the conclusion of his short lecture, as severely "My girl's in Florida," said Will Y at as if one of us had suggested that it.!!!! true, length. "Could that thing write her a poe m "that the Yaffee Electronic Relay Poem-Writer telling her to take it easy and don't get carviII eventually supersede or obsolete the poet. ried a1.fay by no no-good creep jus t b e c a use What it viII do ees to free the poet from the he's gotta sun-burn and one of them foreign sports cars _ I f mechanical trammels that have hitherto shackled him, increase his output, and en han c e his leisure." "Splendid," said the Doctor approvingly. "A popular sobj ect. Let us see which of 0 u r We were then escorted into an adj oini n g categories viII most closely approximate the room, brilliantly lit, lined on all four walls sentiments you have expressed." from floor to ceiling with filing cabi net s, and furnished with business-like looking chairs He crossed the room to the "L's. "'Lover, and tables. In the middle of the room s too d the''', he read. fIVe have broken this d 0 Vi n a squat black object, rather resembling a kitinto a number of sub-classifications. I can chen stove, which, we were inforlOOd, was the offer you 'Lover, the Dejected'; 'Lover, th e . feed, or input unit, of YERP. Dr. Yaffee asked Desolate'; 'Lover, the Desperate'; 'Lover,the for a volunteer demonstrator. Despondent'" -- "I would prefer someone with neither a scientific nor a literary background, n he said. "Thus ve viII get a true picture of the machine's extraordinary capabilities." After some hesitation the young reporter from the news weekly offered himself up onthe altar of science. His name, he told us, VI as Willy. Ordinarily he was on the Sports Desk, but he was at present subbing for a sick friend at the adjacent Science Desk. "First off," Dr. Yaffee instructed Willy, "ve must select our sobject. Please to notice the filing cabinets around the room. T he r e are more than 500 of them. Each drawer has its own title, indicating its sobject, and in the drawers are kept, punched on cards, 0 u r machine's vocabulaty -- one vord to each card, along with various coding punches for the purpose.of identification and recognition by the machine. Here, as you see," said the Doctor, crossing to the far side of the room, "vestart with the sobject 'Aberration'; so on to 'Abnegation', 'Ambition', 'Anger', 'Anxiety',and so forth; on through 'Beauty', 'B r a very' " 'Brutali ty' - I pick at random - rig h ton down through the alphabet to l.zoomorphism' • My colleague, Professor Morgansen of the English Department, ees responsible for the selec- Willy shook his head. said hoarsely. "Just noivous", he " 'Lover, the Apprehens i ve ' ", anno u n c e d the Doctor~ "I think this ees just whatve're looking for." He removed three long, shallow drawer s from the file, in which thousands of 0 b long cards lay neatly stacked: placed ·them on a low table, and wheeled them across the roo m t 0 YERP'S input unit. He then inserted the contents of all three drawers into the machine, briskly tapping the edges. of the s t a c k s to bring them into perfect alignment. When a 11 of the cards were thus stowed, he touched a s1.fi tch. The machine went on with a low hum, and instantly the cards disappeared 1.dthin its capacious maw. "You may think of this unit as the machine's memory," he informed us, "in which are stored these thousands of vords, punched on cards, just as a man stores his vocabulary in his memory. So. Now. Next. Ve must tell the machine what form of verse ve wish towrite, and select our rhyme pattern." "Rhyme!" murmured one of the New J e rse y poets. ''How quaint!" (continued on page 3R) - 13 - I.R.E. NATIONAL CONVENTION, MARCH, 1956, NEW YORKTITLES AND ABSTRACTS OF PAPERS BEARING ON COMPUTERS AND AUTOMATION The Program of Technical Sessions of the IRE National Convention in New York, March, 1956, contains many papers having some relation to computers and automation. Folloldng are the titles and abstracts of 31 of these papers, and notation of the part of the IRE Convention Record in lvhich they will be published. SESSION II mitted over a radio channel of a given width has a close relationship to the complexity of the apparatus involved in its transmission and reception. Virtually every radio service licensed by the Federal Communications Commdssion utilizes far more radio spectrum than needed to convey the necessazy intelligence in order to utilize low cost apparatus. There are excellent expansion possibilities in the land mobile services which are today the most inefficient users of our radio spectrum. Sponsored by the Professional Group on Medical Electronics. To be published in Part 9 of the IRE Convention Record. SESSION IV Sponsored by the Professional Group on CO~ munications Systems. To be published in Part 8 a the IRE Convention Record. Medical Electronics I 2.3 The Application of Automatic, HighSpeed Measurement Techniques to Cytology General Communications Systems W. E. Tolles, R. C. Bostrom, and H. S. Sawyer, Airborne Instruments Lab., Inc., Mineola, N.Y. 4.1 The Cytoanalyzer, an instrument being developed for high-speed, automatic screening of cytological smears for the early detection of cancer, is based on the distinguishable differences that exist between malignant and normal cells when compared ldth respect to nucleus size and density. This paper disc usses the techniques us ed to obtain these differences in cell characteristics from the electrical analog of the smear. The design and operation of the measuring and computing circuits used in the instrument, including evaluation test methods and results, are described. ~le design of the scanning element used to convert the optical info~mation of the smear to a serial electrical current is summarized. Sponsored by the Professional Group on Vehicular Communications. To be published in Part 8 of the IRE Convention Record. Vehicular Communications: "Nelv Horizons for Vehicular Communications" 3.3 A.' O. Mann, SKF Industries, Inc., Philadelphia, Pa. The current status of teletypewriter and related communications equipment for integrated data processing at SKF lvill be described. Further description will be given of the futu~e plans for provision of a complete, national circui try of teletypewri ter. The relationship of such communic ations equipment to a complete computational and control program wi.!l be outlined, highlighting our plans for a complete closed circuitry with full feedback. Included in the paper will be descriptions of some new and decidedly novel communications equipment t'lhich l'le have developed in collaboration ldth A.T. & T. Co. and t~ich exists mwhete else. 4.3 SESSION III More Words Per Minute Per Kilocycle C. B. Plummer, Federal Communications Commission, Washington, D. C. The amount of information lvhich may be trans- 14 - The Place of Communications in Integrated Data Processing Sixteen Channel Time Division Multiplex System Employing Transistors and Magnetic Core Memory Circuits J. C. Myrick, Rixon Electronics, Inc., Silver Spring, Md. and Walter E. Mo rro1v, M. 1. T., Cambridge, Mass ~ A four-channel time division multiplex system, utilizing vacuum tubes, has been in use for several years. This paper describes a new development, lvhich compresses sixteen standard 60 0 r 100 lvord per minute teletype inputs into a tilE division multiplex system developing an output suitabl~or use with frequency-shift keying systems. [be computer field has been drawn on for ferrite core memory circuits, shift registers, Computers and Automation binary count-dotvn circuits, and applications 0 f transistors. The equipment to be described occupies the same rack space, requires far less power input, and is inherently much more reliable than the multiplex equipment currently in use. 3) An important feature of this equipment is the incorporation of timing facilities based on an oscillator with an inherent stability of one part in ten to the eighth per day or better. This provid~ highly synchronous operation with infrequent synchronizing pulses. TIle probability that a Rayleigh distributed random variable tdll fade bel 0 tV a threshold for a given time. Results obtained for the last two functions will be presented. They agree very well wit h known analytical results. SESSION X Sponsored by the Professional Group on Automatic Control. To be published in Part 4 of the IRE Convention Record. SESSION VII Sponsored by the Professional Group on Information Theory. To be published in Part 4 of the IRE Convention Record. Automatic Control 10.5 Information Theory I 7.1 C. H. Doersam, Jr., Doerco-Consultants, Port Washington, N.Y. Information Theory and Quality Control Jerome Rothstein, Signal Corps Engineering Labs, Fort Monmouth, N.J. A basic analogy is described between a communication system and a manufacturing system with the follotdng correspondences between terms: me ssage source and specification, transmitter and means for modifying ratv materials, channel and objects possessing measurable characteristics relevant to specifications, source of noise and cause for rejection, receiver and quality measurement system,' ensemble of received messages and lot of manufactured articles of measured statisticalqumity. The common logical basis of statistical communication theory and statistical quality control, plus the fact that measurement can also be described as communication, assumes particular importance if automation is extended to encompass both quality control and proQuction. 7.5 The "Reasonableness Check" in Automation The nature of automation with respect to the automatic control of a physical process is revieT.ved. The boundaries of extent and rate \\hich limit the physical processes are noted. The new concept of "reasonableness concept" is defined in terms of these boundaries. ExamplE; are given which show that in its most elemental form the reasonableness check is common. Its power when extended to more complicated control situations is discussed. An example of one such problem is given. This example uses a digital computer in an automatic control problem. It serves to indicate some of the methods tvhich have been developed from the basic concept. SESSION XI Sponsored by the Professional ~roup on Aeronautical and Navigational Electronics. To be published in Part 8 of the IRE Convention Record. Evaluation of Complex Statistical Functions by an Analog Computer Air Traffic Control R. R. Favreau and H. Low, Princeton Computation Co., Princeton, N. J. and I. Pfeffer, The Ramo-Wooldridge Corp., Los Angeles, Calif. 11.1 This paper presents a technique for experimentally determining a number of statistical functions which are difficult or impossible to evaluate analytically. Technique developed tfill be described by illustrating its use in evaluating three such functions listed below: 1) 2) The probability distribution'of time to first passage across a threshold for a Gaussian Random variable with a given spectrum. The probability distribution ft> r the length of interval between two successive zeros of a Gaussian random variable with a given spectrum. - 15 - Symbolic Display System for Air Traffic Control L. T. Harris, Griffiss Air Force Base, Rome, N. Y. A general statement is made concerning the work that has been accomplished to date in the air traffic control area, the inadequateness of present day air traffic control equipment, and a revietv of various technical developments that hold promise of being effectively used in air traffic display systems. A plan is presented for a proposed integrated display system capable of providing a nonambiguous display of aircraft identity and position coordinates suitable for high density air traffic control application. 11.2 Computers amI 4u tomation A Nel\' Look at Requirements fl~r Electronic Systems in Air Traffic Control R. S. Grubmeyer, Franklin Institute, Philadelphia, Pa. As new equipments and concepts have been developed for the control of air traffic, simulation and other tests have thrOtvn additional light on the detailed requirements for nelV electronic systems and equipments. Some of these requirements have been met, but others continue to present a challenge to the e1ec tronic industry. This paper 1\'111 highlight the most pressing current requirements, presenting the background information on their development so that alternative solutions may suggest themselves. The primary purpose of the paper is to develop in the industry an increased alVareness of current needs so that research and development programs can be guided along· the most productive lines. 11.3 Traffic Control Electronics Research Goes Modern E. N. Storrs and J. L. Ryerson, Griffiss Air Force Base, Rome, N. Y. This paper describes the system's research and development program of the intercenter Traffic Control Approach and Landing (TRACAL) team of the Air Research and Development Command of the U. S. Air Force. The basic system engineering concepts of this group are outlined in schematic form and the plans for the Phase I, II, and III t r a f f i c control systems are described. Details of the techniques being applied in the development of enroute traffic control, approach, landing, airfield guidance, and display are d i scussed. 11.4 SESSION XVII Sponsored by the Professional Group on Reliability and Quality Control. To be publishe din Part 6 of the IRE Convention Record. Quality Control and Reliability Studies of Electronic Equipments 17.2 Some Reliability Aspects of Systems Design Fred Moskowitz and J. B. Mclean, Griffiss Air Force Base, Rome, N. Y. This report uses elementary princ i p 1 e s 0 f probabili ty theory and a systematic developm e n t is presented which leads to formulas, charts, and guide rules for engineers involved in the design of systems and equipments. Examples are g i ve n which illustrate the use of the formulas and th e principles derived. This study attempts to show that 1" hen the problem is present of obtaining reliable e~ment which consists of unreliable parts, the solutio n is redundancy. Complexi ty by itself need not necessarily lead to unreliability if complexi t y i s used correc tly • 1\'0 very s imp1e redundancy schemes are described and analyzed. It is shotfn t hat ~ t i s possible to obtain a desired reliability at relatively reasonable cost in terms of increased size and Iveight. SESSION XVIII Sponsored by the Profess ional Group 0 n N uclear Science. To be published in Part 9 of th e IRE Convention Record. An analysis for Human Flight Control Nuclear Instrumentation L. J. Fogel, Stavid Engineering, Inc., Plainfield, N. J. 18.2 A mathematical model of some aspects of the aircraft information transfer process is suggested l\'hich includes some usually disregarded human operator characteristics, such as anticipation, amplitude quantization and sequential sampling. Various measures for system performanceevaluation are suggested. These may be used to examine the nonstationary probability density distributwn of the output-message lvith respect to the inputsignal probability density distribution as a function of time. The output-message is def in e d as the actual flight path, .while the input-signal is taken to be the "intended" airpath-that path described by the probability function 0 b t a i ne d from all previous successful performance 0 f the mission phase under consideration. The formulated solution permi ts both nonlinear and time-varyin g elements, provided their transduction rem a ins' single-valued ld th time limited memory. This paper' presents a general survey of the field and an engineering approach to many highly complex displaycontrol design problems. - 16 - Punch Card Recording and Multiple Counting Data H. D. leVine and Henry Sadowski, U. S. Atomic Energy Commdssion, Health and Safety Laboratory, Net" York, N. Y. The system will process data from as man y as 100 counting systems by channeling a complete set of information on a given sample into a central IBM card punch. Automatic interrogation of individual counters permits the elimination of manual techniques and the avoidance of the human erro r factor. Each punch card carries detailed information on the number of the sample, the character of the sample, activity, counting geometry of the counti~g system, counting time, and other related data. Most circuits were redesigned to eliminate vacuum tubes and apply transistor and glow counter techniques. SESSION XXIV Computer1'o nod AutomRtion Sponsored jointly by the Professional Groups on Antennas and Propagation. Telemetry and Remote Control, and Mili tary Elec tronics. To be Published in Part 1 of the IRE Convention Record. Symposium: The U. S. Earth Satellite Program -- Vanguard of Outer Space Chairman: W. R. G. Baker, General Electric Co., Syracuse, N. Y. photo panel and brown recorder and also the use of magnetic tape for recording of high frequency information in fm form to replace the oscillographs now being .used. Both !iystems will allm'\' the use of automatic techniques for processing the d a t a, since the information is recorded in electrically retrievable forms. The lot'l frequency system has better accuracy than present systems and for most cases the high frequency system has the equivalent or better accuracy. 28.2 The prospect of man-made Earth Satellites to be launched in the International Geophysical Year (1957-8) has excited the imagination of engineer, scientist and layman. The launching, placing i n orbi t, cons truc tion of the rockets and satell i t e itself present interesting engineering challenges. Few realize, hotvever, the scope of the prob 1 ems associated tvi th the conununication and colI ec tio n of data from such a missile. The objectives of the satellite program and the scientific gains to be achieved will becovered in this discussion. The major emphasis, however, will be placed upon problems of: 1) Keeping track of the missile which is to be done by radio and, more precisely, by optics, and 2) Gathering: data from the missile. Al 1 0 f this involves radio transmission, propagation and intermittent reception at many points widely separated along the ground, and rapid computa t ions based on such data. The establishment of a satellite might co ncei vably be accomplished without. the use 0 f t·h e electronic art. The use of electronics, however, will increase immeasurably its value to man. Airborne Data Acquisition System W. H. Foster, Electronic Engineering Co., Los Angeles, Calif. The contents of this paper are comprised 0 f the results of Phase II of Project DATml, awarded by EAFB to EECo of Calif. DATUM is the code name for Data Acquisition and Transmiss ion by Unif 0 r m Methods. Phase II is the airborne data acquisition portion of the project. It consists of both accumulati~g and recording flight data. In addi tion to the entire system itself, several new units discussed in the paper are: the airborne magnetic tape recorder, recently developed strain g·ag~ oscillators, the calibration sys t em, and possibly a new transducer to record total fuel· used. This new approach to the accumulation of airborne data by uniform means facilitates rap i d , sometimes "instantaneous" data reduction. I n addi tion, there exis t no problems of time and event correlation, such as existed when some data tV a s recorded on photo panel, some on oscillograph recording, and some on magnetic tape after air to gnd telemetering. With this relatively new system accuracies of 10 per cent are "readily" obtainable. Preliminary checkout of the system, u n d e r simulated conditions, indicates that all des i g n goals have been met, some superseded. Co mpIe t e flight tests will be completed in December. SESSION XXVIII 28.3 Sponsored by the Professional Group on Telemetry and Remote Control. To be published in Part 1 of the IRE Convention Record. F. K. Williams, Rocketdyne Field Lab., Rocketdyne, Canoga Park, Calif. Flight Data Reduction Systems 28.1 Requirements of a High Speed, High Quantity, All-Electronic Data Processing System Handling large quantities of data taken over relatively wide bandwidths is customarily done by hand or electromechanical semi-automatic systems. To circumvent the problem of handling this da t a, Rocketdyne has developed an all electronic, h ig h speed high quantity data system. This system operates on a total bandl.ddth of 1,500 cps (b a sed on Hartley IS criterion) or 10 thousand conversions/ second of nine bits each. Analog information is received from one hundred separate input channels, multiplexed, clamped and converted. This data is recorded in a permanent storage on magnetic tap e capable of holding eight minutes of informationor four million eight hundred thousand, eighteen bit words. Each lyord contains the information producai from a channel and the identification of the channel plus a gross error marker. Since all of the data An Improved System for Collecting and Processing Flight Test Data H. W. Royce, Glenn L. Martin Co., Baltimore, Md. This report outlines a system l'lhich tvill be capable of collecting aircraft and missile flight tes t data and of resolving data reduc tion pro blems presently encountered. At the s arne time this system offers a method for preserving better accuracy and permitting some simplification in adding netf measurements foum\, to be necessary 1 ate in the program. Reasons are presented for the use of digital recording on magnetic tape as replacement for the - 17 - Computers and Automation is recorded in digi tal form on tape, it can 00 usoo to supply information to a digital electronic conr puter, in this case an IBM 701 or IBM 704. Th e taped ral~ data is played back into a data selector circui t tIThich eliminates all unwanted data tim etvise. Finally, the data ,is transcribed onto t t" 0 IBM 727 tape units in blocks of arbitrary length. This blocked data can then be processed directly from the console of the IBM computer in any tV a y desired. Analog records can be reproduced, computing can be done, or punched card or typed data taken from the machines output. The latter part of the system can also be fed from a digi tal radio telemetering system. 28.4 The mUltiplier consists of a simple electronic integrator, a comparator, tt'lO output s~ gates, and an additional comparator for each in pu.t • A five input multiplier is described which operates at a sampling rate of 400 cps tdth a transitio n time to the net" produc t val ue of 100," sec. 32.2 Analog Multiplying Circuits Using Switching Transistors Kan Chen and R. o. Decker, Westinghouse Electric Corp., East Pittsburgh, Pa. Techniques for a High Speed, High Quantity All-Electronic Data Processing System, IDIOT II M. L. Klein, Rocketdyne Field Lab., Rocketdyne, Canoga Park, Calif. The design of a high speed, all-elec tr 0 n i c data handling system requires the use of sever a 1 novel techniques. Multiplexing is accomplis he d with anelectr~-mechanical, mercury jet stvitc h which simultaneously acts as a keying system fo r the whole record. Each input is s 1 amped i n a n all-electronic system which allows a finite period for the conversion to binary code. The convertor, a prograrruned trial vol tage encoder tvhich successi vely trys binary vol tages and executes a fix e d logic, yields a straight binary code output. This output, along with the channel identification and error marker are transcribed onto tape in blocks of six and timing markers added. This tape record is the permanent data storage. To feed the d a t a into a computer, the data is first played in t 0 a time filter t1hich examines only wanted data. Each block of six bi ts is examined for oddness and evenness and a pari ty check mark added to ma i n t a i n oddness of bits. Finally this data is rec 0 rded on two IBM 727 tape uni ts, blocked out into p r eset lengths with ten millisecond gaps i nse r te d wi thout loss of data. This technique makes use of the displaced time head method for keying. Wit h the data available in this form, it can be use d by the IBM 701 or 704 computer from console c 0 ntroL Several million t\Tords of data can be handled automatically in this manner and processed at extremely high speeds. SESSION XXXII Sponsored by the Professional Group on Electronic Computers. To be published in Part 4 0 f the IRE Convention Record. Analog mUltiplication schemes based 0 II the principle of modulated rectangular pulses have been developed using switching transistors and squareloop magnetic cores. A two quadrant mUltiplying circuit employs amplitude and frequency modulatiDn. A four quadrant multiplying circui t employs amplitude and pulsewidth modulation. Each circuit has a high degree of r~produci bili ty and basic simplici ty that is not found i n most vacuum tube multipliers. The accuracy 0 ve r a two decade range of output is as good as t hat achieved by more complex vacuum tube c irc u its. Good temperature stability is possible because the. transistors operate in a SId tching mode. The response time of both multiplying circuits is equal to one cycle of the modulated rectangular pulses. With distinct durability, dependability and long 1 ife, these cireui ts should find tdde accept a nc e in both industrial and military applications. 32.3 A Multiple Input Analog Multiplier Logic Design of the RCA Bizmac Computer A. D. Beard, L. S. Bensky, D. L. Nettleton, and G. E. Poorte, Radio Corporation of America, Camden, N.J. The RCA Bizmac computer has been deve 1 <> p e d as a maj or element of the Bizmac sys tern, and may be described as a general-purpose three-addres s stored-program machine. It has certain specialized features which make it adept in cyclical accounting applications: completely variable word length in all internal operations; highly-flexible i nstruction complement directed toward data-orga~ ing abili ty; a control philosophy which of fer s great operational flexibility and s imp 1 if i e s troubleshooting and maintenance. The present paper tfill outline the control and organiZational concepts of the computer. 32.4 Electric Computers I 32.1 a number of input variables. Positive vo 1 tag e analogs of the input factors are periodically Sampled to produce an output product which changes in discrete steps at the sampling rate. Input and Output Devices in the RCA Bizmac System J. A. Brustman, K. L. Chien, C. I. Cole, and D. Flechtner, Radio Corporation of America, Camden, N.J. D. D. Porter and A. S. Robinson, Columbia University, Net" York, N.Y. This paper will describe the f u nc t ion a 1 characteristics and some of the design feature s This paper describes an electronic a n a log computing technique for obtaining the product of - 18 - of the follotdng equipments: Computers and Automation Tape1.vri ter - A manual keyboard device which creates punched paper tape. Tapewriter-Verifier -- Permits a character-Qycharacter verification of a previ ously prepared tape. Paper Tape Transcriber - Transfers information from the punched paper tape to magnetic tape. Card Transcriber -- Trans lates informa t ion from punched cards to Bizmac code on magnetic tape. Electro-Mechanical Printer - The major highspeed output printer of the Bizmac System. Magnetic Tape Transcriber -- Transfersinformation from magnetic tape to punched paper tape in the RCA Bizmac code. Interrogation Unit -- Permi ts direc t acce s s to the Tape File for a rush random interrogation. 32.5 Burroughs Series G High Speed Printer E. M. DiGiulio, Control Instrument Co., Inc., Brooklyn, N. Y. The Burroughs Series G high speed' printer is a device capable of printing 900 lines per minute from punched cards. It represents the gre ate s t single advance yet achieved in increasing the speed of tabulating and printing machines. This p ape r discusses some of the basic design features th a t make this high speed operation feasible. Chi e f among these are the dual card feed with its independent picker knife control, the unique tiT i r e printing arrangement and the electronic circuitry used for decoding and encoding information to be printed. The paper will also cover some 0 f the features to be incorporated in subsequent machines of this series, such as accumulation, magneti c core storage, a bill feed printer and a pr in t e r punch. SESSION XXXV Sponsored by the Professional Group on P r 0duction Techniques. To be published in Part 6 of the IRE Convention Record. Design Approaches with Printed Wiring 35 .. 2 Principles of Circuit Design for Automation H. S. Dordick, Radio Corporation of America, Camden, N. J. analysis. The technique is applied to a v ari e t y of products and the resultant standardized aut 0mation package is shown. Slides will be presented. SESSION XXXIX Sponsored by the Professional Group on Electronic Computers. To be published in Part 4 of the IRE Convention Record. Electronic Computers -- II Chairman: John H. Hmvard, Burroughs Corp., Paoli, Pa. 39.1 B. Cox and J. Goldberg, Stanford Research Inst., Menlo Park, Calif. A recently announced electronic accounting machine (ERMA) is required to file 50,000 items a day to magnetic tape storage. The filing proceeds continuously during the day and utilizes a magnetic drum as a temporary storage device. Each item is identified by an index number; the items are entered to the machine sequentially in random index number, but are stored on magnetic tape in numerical order. The sorting operation occurs between drum and tape and is accomplished by a unique electronic-sorter, which is characterized by its ability to scan a large number of drum tracks simultaneously td th but a single index number regis ter and a minimum of associated logical circuitry. The system is further characterized by a small number of lvri ting-erasing operations per item. 39.2 A Magnetic Drum Extension to the Gamma 3 Computer P.L. Dreyfus, H.G. Feissel, and B.M. Leclerc, Compagnie Des Machines Bull., Paris, France The BULL Gamma 3, a production line computer, t'las primarily designed tv! th a ,small internal storage to be connected to standard punched card machines. An extension including a magnetic drum and high speed storage may not\' be connected to the existing model, increasing a thousand fold its internal storage. This paper will describe logical and technological problems.involved in this connection and some basic features of drum circuitry. 39.3 The equivalence of circuit design,requi rements for high volure automation and j ob- s ho p automation is shotvn. A technique of a n a I y sis known as sub-modularization is described. Th i s resul ts in circuit elements of standard s i z e , content, configuration, and manufacturing p r 0cessing. These elements are applicable to many diverse types of equipment, creating a mass produced type of product within the, job-shop. A mathematical representation is given which aids in standardization of circuits and systematizing the A Magnetic Drum Sorting System The Univac Magnetic Computer - Part 1. Logical Design and Specifications A.J. Gehring, L.W. Stowe, and L.D. Wilson, Remington Rand Univac Division of Sperry Rand Corp., Philadelphia~ Pa. This paper describes a two address, decimal serial, binary parallel computer which uses about 1,500 magnetic core devic~s together with germanium diodes to perform all arithmetic and control functions. The arithmetic element lIS:"S four rna/]_ 19 (continued o~ page ~2) Free an'd Use the of the Remote Toronto Computer, Programming of It PART 1 ~ 1. c. GOTLJER and'others Computation ~entre University of Toronto Toronto, Canada Note By the Editor: III. In the December 1955 issue of "Computers and Automation", The Editor's Notes contained a brief report of a conversation in Boston lvith Dr. C.C. Gotlieb of the Computation Centre at the University of Toronto. Dr. Gotlieb said that they had a policy' of allowing free use of their Ferranti computer Ferut, to a reasonable extent, to any investigator (t"hether in Canada or not) t"ho was not going to profit personally from the research: this in spite of the fact that they regularly charge $100 an hour for the use of their computer on commercial problems. He invited the putting of problems of this class on Ferut. PREFACE This manual was written for scientists, engineers and others in Canada to make available to them the use of FERUT, the automatic electronic digital computing machine at the Computation Centre in the University of Toronto. It is intended also to acquaint other readers with the systems used here. With its aid one can write programs for computations by the machine without having to learn the many intricacies that must be mastered by the professional programmer whose concern is with the efficient use of the machine in long calculations. Already programs in Transcode have been written outside Toronto and the results of computation sent by mail. Wherever a number of similar calculations have to be done as part of a research, it is advantageous to use this method in place of a desk calculator. For calculaiions that do not involve many hours of machine time, Transcode overcomes the difficulty of providing sufficient expert prograffimdng help to accomplish all the t"ork for tvhich the machine is sui table and for which computing time is available. In addition Dr. Gotlieb said that they were planning to tie in their computer by teletype with many other universities in Canada, so that each co~ld have access to the machine and put its ot"n problems first hand onto the machine. In this way they could make their one machine helpful to the whole of Canada for research and instruction. II. From "Transcode Manual" -- ~_System of Automatic Programming for FERUT, the Ferranti Mark I Electronic Digital Computer at the University of Toronto, published by the Computation Centre, University of Toronto, Canada, October, 1955, 58 pages: From Dr. C.C. Gotlieb: I thought that was a very fair report of the conversation t"e had in Boston. The programming of our machine is such that we have not exactly been swamped by requests from outside users to avail themselves of the free machine time for "r.eal code" problems. The general ideas embodied in this system were evolved by Professors Hume and Gotlieb. The very substantial amount of expert programmdng required to prepare the machine to translate the simplified Transcode program was carried out by Professor Hume and Dr. Worsley. The system was 'first put in operation in September, 1954. Since then a number of elegant improvements, especially those due to Dr. Kates, Dr. Worsley and Mr. Watson, have been incorporated. Much ingenuity has gone into the routines for the machine that the user of Transcode unconsciously relies on. Prof. Griffith and Mr. Weir have also contributed library routines. The manual was prepared by ~ fessor Gotlieb, Dr. Worsley and Mrs .. A. Wallis. Already over fifty persons have used the system in their work. To encourage learners, it may be mentioned that graduate students in science and engineering at Toronto have learned readily in a very short time and are enthusiastic users. I am sending under separate cover a cop Y of our new Transcode Manual which contains a description of our system of automatic programming for Ferut. We find we can teach this system in a few hours and Transcode programs from outside Toronto have been coming in steadily. You will also be interested to know that we have had five evenings of extremely successful runs tvi th the Uni versi ty of Saskatchelvan via the teletype link supplied by the Canadian National Railways. It D;lay be that CNR will provide us wi th free tel etype lines to any Canadian uni vers i ty and the experience t"e have had to date makes us extremely enthusiastic about this way of running. There are already some fifteen people at the University of Saskatchetvan who have run problems on Fer u t wi thout ever having seen the machine, and w e feel that we can appreciably increase the number of students tiho acquire experience on a computer during their university career. For those without much knowledge of computing by machine who approach this manual hopefully, may I suggest that the main initial hurdle to be overcome is the mastery of new vocabulary. A serious attempt has been made to make the reading - 40 - ComputerN Rnd Automlltioll of this manual easy in that respect for them. At the same time it is fair.to point out that although FERUT operated Iii th Tra:tscode accompl ishes in one hour as much calculation as a man with a desk machine can do in one month, the l'lri ting of a program means producing in some detail the series of instructions required. Accordingly attention to detail cannot be avoided. HOl1ever it is soon 1earned and the aide memoire or summary sheet requires only a single sheet of paper. Chapter 4. Examples ••••••••••••••••••••••••••• 33 Chapter 5. Libra~y Appendix Lest it should be thought that Transcode is a sort of programmdng only for amateurs I should like to mention that in the interest of speed of computing FERUT is a fixed-point machine ordinarily. Ho'tyever, programming tv! th fixed dec i mal point encounters on occasion quite exacting problems of scaling. By its use of floating point, Transcode is a valuable aid to the programmer facing scaling problems and indeed in some calcu1.ations Transcode is more economical than fixed point computation. Since, in Transcode, numbers are submitted to the machine in decimal form, the user is saved all contact with binary arithmetic. Finally in introducing the reader to the authors may I suggest that he keep in mind that he is investing his time in a rewarding venture. Hours of desk computing are handled painlessly by the machine in a few minutes. As Director of the Computation Centre I moWd like to take this opportunity of expressing my appreciation of the enthusiasm and hard work of the Centre staff in creating this new facility at Toronto to serve the interests of scientific co~ puting in Canada. I hope their efforts will be rewarded by a good response from the physicists and others it is written for. III IV V 1.1 GENERAL Electronic Digital Computers The FERUT computer is one of a type generally described by the terms digital and auto rna ti c. The term digital is used to distinguish such co~ puters from analog, or continuosly-variable d evices, of which the slide-rule and differenti a 1 analyser are common examples. The earlier digital calculators, such as the abacus and desk-type adding machines, lyere designed to mechanize only the ari thmetical operations which occur in han d calculations. H01vever, these r~quire a human operator to carry out such processes as in pu tti n g data, transferring intermediate results from one register to another, transcribing the final anStfers into a presentable form and possibly e~loy ing judgment at some stage in the calculation to decide between alternative procedures. The term automatic is applied to computers which are des igned to carry out these processes wi thout the intervention of an operator. FERUT can also be described as a stared-program calculator. That is to say, it functions by obeying a sequence of instructions which must first be stored within the machine in some coded form. TRANS CODE MANUAL TABLE OF CONTENTS 1.2 Functional Units Computers satisfying the above description are now named after Dr. A. M. Turing of Mandeste~ who first postulated the theoretical form of a machine t.vhich would calculate any computable nu~ ber. In practical terms, Turing machines must contain five basic units: Chapter 1. General 1.1 Electronic Digital Computers •••••••••• l 1.2 Functional Units •••••••••••••••••.•••• l 1.3 Machine Instructions ••• ~ •••••••••••••• 2 1.4 Automatic Coding Techniques ••••••••••• 4 Chapter 2. The Transcode Machine 2.1 Machine Description ••••••••••••••••••• 6 2.2 Transcode Instructions •••••••••••••••• 9 2.3 Tape Controls •••••••••••••••••••••••• 19 3.1 3.2 3.3 3.4 3.5 3.6 CHAPTER 1. Finally, FERUT may be described as a un i versal calculator in the sense that it can be used to solve any problem which can be reduced to a sequence of numerical operations on given data. Only considerations of speed and storage capacity can qualify this statement. W.H. Watson, Director. Chapt~r Magnitude Restrictions on ~ Transcode Number ••••••••••••• 53 Ferut Operating Sheet for Transcode ••••••••••• 54 Times of Transcode Operations •••• 56 Glossary of Terms •••••••••••••••• 57 Summary Sheet •••.•••••••••••••••• 59 I II If any learner experiences difficulty we should like to hear from his so that lye may profit by learning where lye failed as instructors. Functions ••••.••••••••••••• 43 (i) (ii) (iii) 3. Operating Notes Program Design ••••••••••••••••••••••• 22 Tape Preparation ••••••••••••••••••••• 23 Console Procedures ••••••••••••••••••• 25 Write Taping ••••• ~ ••••••••••••••••••• 26 Transcode Zero and Infinity •••••••••• 27 Special Coding Techniques •••••••••••• 28 (iv) (v) an input unit, a set of storage registers, an arithmetical unit capable of performing logical or arithmetic operations on arrays of dig its stored in the registers, an output unit, and a control unit which arranges for the functioning of the computer as a whole. The set of available operations constitutes - 21 '- Computers and Automation the instruction code of a given computer. The program, or set of these instructions r~quired tosriWe a given pt:oblem, must be prepared by a proc es s known as coding, and fed into the store 0 f the comput~r before being obeyed. The individual instructions then assume the same physical fo r m as numbers within the machine. The operation of computing involves the following steps: (i) (ii) (iii) inputting the program, along with any necessary data, such as starting values, parameters, and tables of empirical functions, initiating the calculation by causing the control to proceed to the first instruction, removing the final results, (outputting instructions are generally included in the program). A machine like FERUT makes use of special storage registers to carry out its operations. There is an accumulator, a register into which the immediate re,sul ts of an arithmetic or logical operation are placed. ,There is also a set of B-registers which can be used to modify machine instructions as they are obeyed without altering their form in the store and without recourse to the accumulator. B-registers can be used as counters, since addition and subtraction can be carried out directly on their contents. Digit representation within a computer need not be decimal. Scales of 32, 8 and 2 are common. The binary, or scale of 2, representation is used in FERUT. Each bit, or unit of binary information, may be represented by an on-or-off device, and recognized by the presence or absence of a b rig h t spot on a cathode-ray tube. 1.3 Machine Instructions A machine instruction consists of two part~ a functional operator and an operand. The operand generally consists of one or more addresses, that is, names of store locations which contain relevant information. Thus a typical single-address instruction might be ADD 567 Th is l\1ould ins truc t the computer to add the number stored in the 567th storage location to the number already contained in the accumulator. A three-address instruction could take on the form: In FERUT, instructions are single-address. Each is represented by one "line", i.e. a set of '20 binary digits (bits) of stored information. Of these 20 bits,' the last 6 are used to specify the operation, the'first 10 to specify the address of the operand, 'and the remainder to specify the B-register used to modify the instruction. A B-register contains the same number of bits as a stored line, and is therefore commonly referred to as a B-line. Outside the machine, each instruction is represented by four characters, each 5-bit character being a symbol in the standard teletype code. It should be noted that machine instt:uctions are generally quite elementary. Each one achieves only a fragmentary portion of the entire task to be performed so that much detailed work is necessary to prepare a problem for automatic solution. Another practical difficulty often encountered by the programmer for a universal computer is that of scaling. FERUT is a fixed-point ma chine, that is to say, the binary point ~emains in a fixed position relative to the binary digits used to represent a number within the machine. Hence numbers within a limited t:ange of magnitude only are allowed to occur in the course of a calculation. For example, one may be using the machine in such a lfay that all numbers are required to lie between -1/2 and+l/2. Since it is possible for the sum of a set of numbers to exceed these bounds even though each individual number lies within them, suitable scale factors must be attached to the individual numbers so that such an overflol\' does not occur. Compensation for these factors must be made elsewhere in the program. The problems of scaling become acute when scale factors cannot be chosen so as to permit retention of a sufficient number of significant figures. It may then be preferable to represent numbers in floating form, that is, as standardized numbers or mantissae tv! th appropriate exponents. Thus -736.25: ~7.3625 x 102 and may be repre~ted by the mantissa -7.3625 and the exponent 2. A fixed-point computer can be made to handle floating numbers by the use of special coding methods. Similarly, of course, a binary computer can be used to deal with numbers represented in any other scale of notation. 1.4 SUBT ABC anJ instructs the machine to subtract the number in storage location B from the number in storage location A and place the result in storage location C. Machines in which instructions are not obeyed sequentially must provide for the operand to include the address containing the n ext i nstruction. Thus a two-address code migh.t pr 0 v ide instructions like ADD 0052 in the address 0052 to the accumulator and select the next instruction out of address 7631. Automatic Coding Techniques Experience has shown that certain routines (or sub-sections of programs) can be devised in a sufficiently flexible manner to carryover from one problem to another. Thus routines can be l"ri t ten once and for all to perform such operatiom as: (i) 7631 (ii) which l'Jould require the machine to add the number (iii) - 22 - reading-in instructions through the input unit, converting them into the form required by the computer and storing them as required by the program, similarly inputting decimal data, evaluating certain functions such as C.omputers and Automat101l (iv) (v) (vi) sines and cosines, performing calculations such as advancing the integration of a system of differential equations by one step in accordance with some established numerical method, printing out data, with all necessary conversion from machine form to required layout on the printed page, arranging to perform a sequence of instructions a prescribed number of times before proceeding to the next part of the calculation. This last operation is commonly known as looping, and is often accompanied by a systematic modification of some of the instructions in the s~nce. It is this technique t.vhich really makes a storedprogram calculator workable, since it takes advantage of the repetitive nature of calculations suitable for automatic execution. To use a computer efficiently, it is necessazy to consolidate and unify coding techniques. First of all, a scheme of input and organization of routines must be adopted. Compatible Id th this scheme, a library of routines must be constructed to perform such operations as have just bee~ described. There is still a considerable amount of work to be done for each problem by way of co-ordinating library routines into a completed program and this entails a full knowledge of the code and specifications of the machine. Thus the final step towards automatizing coding techniques is to arrange for the computer itself to do the organizing and coding by a pseudocode. Transcode is an automatic coding systemt~t ten especially for FERUT. It derives its name from the fact that it arranges for all information, prepared for input in a simplified form, to be read into the machine from a punched paper tape and translated into the form required by the computer. This translating tak~s place automatically on c e and for all before the calculation proper is started. Translation routines of this type are a 1 s 0 known as compiling routines or compilers. CHAPTER 2. 2.1 THE TRANSCODE MACH INE Machine Description When programming t.'Ii th Transcode, FERUT may be regarded as another machine in Ivhich the specifications for numerical representation, storage organization and instruction code are quite different from those for the "real" machine. The following are the properties of the Transcode machine. 2.11 Number Representation It is not necessary for the Transcode user to become familiar with the binary system. Numbers normally entering into a calculation are expressed in the floating decimal system, that is every number has associated t"li th ita power of ten which may be considered as its scale factor. For example -107.345 is written -1.07345 x 10 +2. - 23 - Since the normal machine code of FERUT is constructed to operate on numbers in fixed-point binary form, a conversion must take place whenever floating decimal numbers are read in. For in Transcode each floating-decimal number is, fact, represented by its equivalent floating-binary form, two lines (40 bits) of machine storage being used to represent the mantissa, and the fo~ lowing line (20 bits) to represent twice the binary exponent. (The factor two in the exponent was adopted for convenience in coding). It is possible to carry as many as 12 significant decimals in a Transcode calculation. Integers, as well as fractions, can of course be punched in floating-decimal form and read into the machine. They then occupy three lines of machine storage. The representation may not always be exact, however, since an error of 1 in ~O may be introduced by the conversion. When counting iterations it is necessary to represent integers exactly within the computer and of a length which can readily be used in a B-line. Therefore an alternative representation has been provided for integers. All the numbers used in B-lines are 20-bi t integers in fix e dpoint form. Note also that the exponent part of a floating number is an even integer of this sort. 2.12 Storage OrganiZation The Transcode machine, like FERUT, has two levels of stores, electronic and magnetic. The ELECTRONIC store consists of three pages* labelled X, Y and Z. There are 21 addresses or store positions on each of the X and Y pages, labelled XOl, X02, •• X21; YOl, Y02, •• Y21. The Z page contains only thirteen addresses ZOI •• Z13. All of these electronic storage positions are individually and simultaneously available during computation and they may be used to store data, intermediate results or tables. The MAGNETIC store consists of 64**DRUM locations, divided equally into two sets, called the lot..rer and upper halves of the range. Each DRUM location can be used as an auxiliary store for a block of 21 electronic store positions. The contents of the X or Y page may be transferred as a whole, to or from a DRUM location by means of instructions described below. The Z page can be transferred only under the conditions of Section 3.64. Such an operation replaces the previous set of 21 numbers with the new material. DRUM locations may also be filled directly from the input tape. It should be noted that the magnetic store is only accessible in block form, this being an engineering feature of the machine. Transcode also makes provision for storing up to 21 CONSTANTS per program. These are labelled COl, C02, •• C21. They retain their identity 'throughout the execution of a calculation and can only be introduced by input from tape before the calculation commences. They are actually stored among the translated machine instructions to become available as required by the program. * These pages correspond to the cathode ray tubes which can be viewed from the console. Each r...omputers and Automa tion contains 64 lines of machine storage arranged' i n tl\t) col umns. In the X and Y pages, the 1 a s t 63 lines are used for storage of Transcode numbers, and the first is spare. In' the Z page, the 1 as t 39 lines are used for storage of Transcode numbers, and the remainder are used by Transcode for monitoring purposes and must not normally be altered by the program. 2.2 Transcode Instructions 2.21 Arithmetic Instructions The ins truc tions XOl.O XOl.O XOl.O X01.0 ADDN SUBT MULT DIVD ** This number may be increased by special arrangement with the operator. Z01.0 ZOl.O ZOl.O ZOl.O YOLO YOl.O Y01.0 YOLO have the effect of placing respectively in ZOI the four results (XO!) + (YO!), (x0!) - (YO!), (x0!) x (YO!) and (x0!) -:- (YOU. In general, the fir s t two addresses may be any of the X, Y, Z or C 1 0cations, and their contents are left unchanged by the operation. The third address may be an X, Y or Z location. 2.13 Instructions The program for a calculation consists of a consecutively numbered set of instructions. These are numbered in sequence,' as 001, 002, 003 etc. Thus each instruction in a given program can be identified by its "instruction number", jjj. Instructions are of the three address type, and always take the form of: The instruction 1/2QRT a four letter functional part specuying the operation to be done. (ii) three addresses, each having four digits, specifying the storage locations of the operands, results, arguments, etc. XOl.O 000.0 Z01.0 (i) The first three of the four digits of each address speclfy a storage location as d efi ned above. The fourth digit can refer to a B 1 i n e, as described belo1v; for the moment it tfill be taken as zero, which indicates no B-modification. Fo r some instructions, this four-digit address may take on a specialized form, such as an integer, specifying how many Transcode numbers arero be printed. It may even be a dummy, in tfhich case it is represented by four zeros. The exact meaning of e a c h of the four digits in each of the three addresses is defined below for each instruction and thisinformation is summarized in Appendix V. has the effect of placing ""(XO!) in ZOI. The instruction KOMP XOl.O I (XO!) I - places Y01.0 ZOl.O (YO!) in ZOI. The possible addresses for these two instructions are as for the four instructions above. The use of the first five instructions ~ obvious and in fact they are the basic compu tin g instructions. The KOMP (compare) instruction has many uses; in particular if (Yon zero it stores the modulus of a number. This is often necessary in examining for convergence or in. determining if two supposedly equal res ul ts agree (see ex amp 1 e below for calculating ~ - >' ). = A few preliminary remarks may be made abo u t Transcode instructions. "Storage location" her e is to ~an magnetic or electronic. The contents of any" storage location are indeterminate, unt i 1 filled by a programmed operation. The conten t s of storage locations used as arguments in a Transcode operation are left unaltered by the operation. The resul ts of a Transcode oper"ation are pIa c e d in the designated storage location, imp 1 yin g a replacement of information previously s tor e d therein. Constants, however, can only be intr 0duced initially from the input tape. 2.22 Let (x0!) be used to represent the contents of storage location XOI. Then the instruc t ion "add (XO!) to (YO!) and place the result in ZOl" and having the instruction number ooj, is written as OOj ADDN XOl.O YOl.O ZOl.O places floating point zero (here .• ' _10- 10 ,000) in XOI and may be applied to any X, Y or Z address. It is useful for clearing locations bef 0 ret h e start of an iteration. Here the last tlfo addresses are dunmies. The instruction OVER (j) (iv) XOl.O 000.0 ZOl.O transfers (XOl) to ZOl, leaving (XOl) unchanged. It has possible addresses as for the first fo u.r instructions. Note, however, that the second address here is a du~. ZERO XOl.O 000.0 000.0 The next two instructions arrange for the transfer of information between the DRUM and electronic store. The instructions for the TRANS CODE machi n e tdll be discussed under the following headings: (ii) (iii) Transfer Instructions Arithmetic instructions Transfer instructions Looping and control-transfer instruc-' tions Miscellaneous instructions - 24 - READ 001.0 000.0 XOO.O copies the contents of DRUM position 001 0 n to page X, i.e. positions XOI to X21. This is commonly referred to as "reading down from the DRUM". Computers an~ Automation WRlE 001.0 000.0 iOo.o copies the contents of page X on to DRUM position 001. This is commonly referred to as "writing up to the DRUM". In general any of the DRUM positions, and the X or Y page (but.!!Q.1 the Z page) may be use d • These instructions may be B-modified as explained at the end of this section. The analogy of reading pages out of or Il(ri ting pages into a book might be noted. Note that' the address of the last number to b e operated upon is written into the instruction address which is to be progressively mOdified, the fourth digit being the number of the B line co ntrolling the loop. While the above three instructions are the ones written into the program, the instructions actually obeyed are carried 0 u t i n the following sequence: LOOP ZERO TR~ ZERO 2.23 Looping and Control Transfer Instructions TR~ ZERO To achieve econo~ of effort in writing programs, it is necessary to arrange that sequence s of instructions be used over and over again, i.e. to Il(ri te loops. B-lines have several uses, and one' of the most important is to act as coun te r s to facilitate looping. The LOOP and B-conditionru TRNS (transfer control) instructions h a ve bee n specially devised for this pu-rpose. Suppose th a t it is required to cycle through a set of instructions 21 times, before proceeding to the next instruction. This can be achieved by 001 002 LOOP 021.0 000.3 TRNS 002.0 000.3 I~T 000.0 The process of looping is very commonly accompanied by the progressive modification of certain instructions Id thin the loop. Tra ns cod e enables this to be done automatically, using the same B-line that controls the loop. 001 002 ZERO ZERO XOl.O X-2.0 000.0 000.0 000,.0 000.0 021 ZERO X2l.0 000.0 000.0 TR~ 000.0 000.0 Set B 3 to 3(21-1), an integer. Place "zero" in Store X21 + (B3) • Subtract 3 from (B3) and test the sign of B3. If it is positive or zero, send control to instruction 002. If it is negative, procee d to the next instruction, n a mel y 004. In general any of the X, Y, or Z ad dre sse s can be B modified. A control transfer can tak e place to any instruction in the program, and may be made conditional on anyone of the B lines 2, 3, 4, 5 or 6. Note that the addresses of constants can not be B modified. 001.0 000.0 000.0 causes control to obey instruction 001 n ext u ncondi tionally. The last two addresses are dummies. TRNS 001.0 000.0 X01.0 causes a jump of control to 001 if the mant is s a of (XOl)~ 0, othe~~ise the next instruction will follow in the usual l'\(ay. In general, the tit i r d address here may refer to any X, Y or Z positio n but not B modified, and the second address is a dummy. The following method allows a much s h 0 r t e r seq~ence of instructions to do the same thing: 000.3 000.0 000.3 000.0 000.0 Magnetic storage locations can also be p r 0gressively modified.- However, since it is generally required to modify them in ascending and consecutive sequence, a different technique, r equiring tll(O B lines, is necessary. This is d i scussed later. TRNS 021.0 X21.3 002.0 002.0 X21.0 004. There are two other control transferinstructions which are useful. Suppose, for example, it is desired to place ZERO in each of the locations X01, X02, ••• X21 • This could be accomplished by the twenty-one instructions LOOP ZERO 000.0 000.0 000.0 000.0 000.0 000.0 The mechanism by which the above pro c e s s takes place is as follows: numbers in the X, Y or Z addresses are actually stored in reverse sequence Iv! thin the machine, separated by 3 machine address units. During translation, the 3 written instructions become, in words 001. 002. 003. The LOOP instruction here prepares the machine to repeat 21 times the set of instructions which follows, B3 being assigned in this case as the counter for keeping track of the number of the iteration. The third address here is a dum my. The TRNS instruction terminates the sequencetobe cycled through in the following way. It m a k e s the machine select as its next instruction the one numbered 002, provided That B3 has not ye t counted off the numbep of iterations specifiedby the LOOP instruction. When B3 shows the co u n t to be complete, the TRNS instruction allows control to proceed as usual to the next instruction, 00 (n+2) in the numbered sequence. 001 002 003 in 000.3 000.0 000.0 000.0 000.0 000.0 000.0 (Set of INST to be (obeyed 21 times over) OOn 00 (n + 1) TRNS ZERO 021.0 X01.0 002.0 X02.0 002.0 X03.0 This last control transfer instruction may be combined wi th the KOMP ins truc tion to m a k e a simple way of testing when some iterated proces s produces a negligibly small result. Suppose for example it is desired to compute -Q,-I. from its 000.0 000.0 000.0 (continued on page 14) - 25 - WESTERN San Ti tIes JOINT" COMPUTER Francisco, and CONFERENCE, February, 1956 Abstracts of Papers The Joint Computer Conference Committee (formed by the American Institute of E~ trical Engineers, the Association for Computing Machinery, and the Institute of Radio Engineers) held the Ninth Joint Computer Conference in San Francisco, Calif., Feb 7-9, 1956. The proceedings of the conference will be printed soon, and may be purchased from any of th"e sponsoring societies, for example, from the Association for Computing Machinery, 2 East 63 St., New York 21, N.Y. Following are the titles and abstracts of the papers given. Tuesday, February 7 10:00 a.m •• Noon OPENING SESSIOl'1 Terrace Rooni Chairman Oliver Whitby, technical program coordinator, Stanford Research Institute Keynote Speaker Norman H. Taylor, computer systems engineer, Lincolu Labs., M.I.T. Walter E. Larew, brigadier general, chief, Army Communications Service Div., Office of the Chief Signal Officer Harold Silverstein, special assistant to the Chief Sig~al Officer Benedict Jacobelis, captain, Office of the Chief Signal Officer Computation of electronic exploitation techniques for weapons systems and research activities has been a major activity in the military services for over a decade. The past few years have been marked with stimulated interest in the application of these same electronic techniques to the business type activities of the Army. Acting under guidance from the top levels of command the Army has established an aggressive program to outline electronic data processing systems in supply operations, personnel, and fiscal management and other administrative activities. Although this program is aimed at far.reaching improvements through long-range planning and operations research, it is planned to take action in those areas which offer immediate benefits. :! :00 • 5 :00 p.m. PROGRAMMING AND CODING Terrace Room A Truly Automatic Computing System Mandalay Grems, Boeing Airplane Co. R. E. Porter, Boeing Airplane Co. A mathematical computing problem can be given directly to a digital computer as a set of algebraic expressions. These algebraic expressions are written in terms of familiar symbols for parentheses, parameters, constants, arithmetic operations, transcendental functions, and a few logical operations. The computer itself interprets and translates these expressions to machine instructions and automatically records these in· structions in a form which can be used repeatedly with input data for computing results. The input data are entered as decimal coefficients with reference to a parameter or con· stant. The coefficient includes a decimal point and can be accompanied by a "power of ten." The computing is performed in a floating decimal system and no scaling of values is necessary by the originator of the problem. Lincoln Laboratory Utility Program System H. D. Bennington, Lincoln Lab., M.I.T. C. H. Gaudette, Lincoln Lab., M.I.T. This paper discusses a utility program system to assist the coding, checkout, maintenance and documentation of large. scale, control programs. A typical program contains 50,000 instructions, one million bits of data storage, and is pre- . pared by a staff of 20-40 programmers, many of whom are relatively inexperienced. The utility system requires 25,000 registers. An Automatic Supervisor Jor the IBM 702 Bruse Moncrieff, Rand Corporation The operation of a large-scale data processor making ex· tensive use of magnetic tapes is a routine·dominated situ· ation. Recognition of this leads to the hope that many of the operating procedures can be given over to the machine. The motive is to increase the operating efficiency by cutting down the non-productive time between jobs; by reducing the human effort in tape handling and identification; and by reducing the re-run time caused by operator errors. Chairman Francis V. Wagner, group leader, eNgineering computing, North American Aviation, Inc. Gestalt Programming: A New Concept in Automatic Programming Douglas T. Ross, Servo Lab., M.LT. \ Gestalt Programming is a desired special language by which a human and computer can converse with each other. Its function is to allow~ both easily and quickly: (a) the computer to inform the human of the computer's troubles, (b) the human to inquire as to the status of the solution of a prob· lem being solved on the computer, (c) the computer to do a major portion of its own programming, or (d) the solutions of problems involving both human and computer decisions. 2:00 p.m.· 5:00 p.m. AUXILIARY EQUIPMENT Nob Hill Theater Chairman W. F. Gunning, Beckman Instruments, Inc. 26 - Computers and Automation Engineerin!! Desi!!n of a Magnetic Disk Random, Access JUemory Magnetic Recording Head Design A. S. Hoagland, assistant professor, University of California T. Noyes~ project engineer~ I.B.M. W. E. Di('kinson~ project engineer. I.B.l\I. An analysis of the process of magn~tic recording of digital data is presented from which qualitative head design concepts are developed and their usefulness demonstr~ted, both in the evaluation of structures and in design for high density storflge. The vector nature (or three dimensionality) of the problem is considered and the condition of operation is non-contact. The IBM 305 :Magnetic Disk Random Acce:;s File is a 5,000.000 character storage unit with fairly rapid access to any individual record. The random access time places this storage medium in the range between magnetic cores and magnetic tapes. The general construction and layout of the machine will be reviewed. The access mechanism and posi. tioning of the magnetic heads is discussed in detail and a description of the air-heads is presented. Recording densities are variable and a modified non·return·to·zero method of recording is used. A '1 erminalJor Data Transmission Over Telephone Circuits E. B. Ferrell, switching research engineer, Bell Telephone Laboratories "Print" Coding System for the In a recent experiment, a simple terminal for data transmission has been demonstrated. Such a terminal might be associated with a special telephone line. Between two such terminals it would be possible to send data back and forth at the rate of 750 bits per second, or 1000 words per minute. The demonstration equipment involved magnetic tape to magnetic tape transmission using amplitude modulation of a 1200·cycle carrier. It err,ployed a 7·bit self-checking code. 705 U. W. Berner, I.B.M. Corp. PRINT (PRe-edited INTerpretive system) is basically an interpretive system which incorporates a number of compiling features in a pre·edit routine. The cost of interpreting versus compiling has been found to be only five percent greater in time and much less in compactness, e.g., the multiplication of two matrices requires the writing of only six pseudo commands. A complete method of symbolic indexing and diagnostics is incorporated. The system retains the compact· , ness and efficiency of the interpretive method with minimum interpretation time. The Use of the Charactron With An Era 1103 Ben Ferber, supervisor, Digital Computing Lab., Convair The IBM Type 705 Autocoder As an aid in debugging, the Char act ron can display the contents of memory' at the rate of fifty words per second. A floating point program can be traced at the rate of ten commands per second. For problems with many answers required at each interval, the format can be vertical for ease in reading, While the Computer is calculating and displaying one page of answers, the camera is fixing and developing the previous page. The Charactron has also been successfully used as an aid in editing input data. Roy Goldfinger, I.B.M. Corp. This paper describes a system of automatic coding being developed for the IBM Electronic Data Processing Machine Type 705. The Autocoder System permits the programmer to define records and constant data in terms of their English names and character lengths. Ordinary 705 operations, combinations of 705 operations, called macro-operations, and library functions may make reference' to defined fields and records by name. A New lUagnetic Tape Handler for Computer Applications Program Interrupt on the Univac Scientific Computer Robert M. Brunlhaugh, Ampex Electric Corporation The recently announced Ampex Series FR200 magnetic tape transports have been expressly designed for the storage and processing of information in digital form. This Series features 5 millisecond start·or-stop time, single· loop threading, and a unique simplified servo tape feed control. Various elements of the tape transport mechanism are discussed in detail, including the basic design considerations, and operation of the servo system. Typical application'i in the computer field are outlined. 6 :00 p.m. - 8 :00 p.m. IB~I ;. Mersel, Supervisor, 1103 (Univac Scientific Computer) Computing Group, Remington Rand A computational run on a computer involves several different types of operations such as input, output, and special computing routines. These all involve transfer of control in the computer program and in the case of input-output operations, close synchronization is required in the program. Computer programming can be simplified and versatility can be increased by building into the computer certain facilities for automatic transfer of control. COCKTAIL PARTY Venetian Room Wedllesday, February 9 :00 a.m. - Noon MACHINE DESIGN Terrace Room (,~ :00 Chairman William L. Martin, director of research, Marchant Research, Inc. a.m. - Noon SYSTEMS Noh Hill Theater Chairman G. D. McCann, professor, Electrical Engineering, California Institute of Technology Requirements for a Rapid Access Data File A. Pulse Duration Modulated Data Processing System George Eisler, Electronics Division, National Cash Register Co. John Lowe, Douglas Aircraft Co., Inc. Jack Middlekauff, Douglas Aircraft Co., Inc. General purpose data processing maehines now on the market are limited in their performance by the electronic data file systems associated with them. This paper discusses the various means of organizing such file systems with the hope that equipment designers will be helped in achieving a much needed solution. The desirable factors havin~ the greatest influence on system utility are discussed in detail: Speed, addressing, capacity and volatility. PDM telemetry data are recorded on magnetic tape in analog form. Time marks are recorded in a second channel of the same tape. An analog-to·digital converter (Magnavox Series 200) translates this analog tape to a digital (pure binary) tape of a form suitable for reading into an IBM Type 701 EDPM. The 701 performs a number of operations including 2i - Computers and AutomationElectrical Engineering, Vniversity of California integrating the time marks, edItmg and checking, stripping, scaling, and calibrating. It punches binary cards for plotting on an IBM Type 407 Accounting Machine. An Experimental Monitoring Routine for the A P.D.M. Data Converter Helen V. l\leek, Programming and Operations Research w. R. Arsenault, project engineer, Magnavox Research Staff, Hughes Laboratories The Logical Design of a Digital Computer for a Large Scale Real-Time Application An Improved Multichannel Drift-Stabilization System 1\1. 1\1. Astrahan, B. Housman, and W. H. Tholnas~ I.B.M. J. F. Jacobs and R. P. l\layer, Lincoln Lab, l\I.I.T. Peter G. Pantazelos, research engineer, Massachusetts Institute of Technology Drift stabilization is essential in doc amplifiers in electronic differential analyzers. To make possible the simultaneous drift stabilization of 30 computing amplifiers, a multichannel drift·stabilization system was built at the Dynamic Analysis and Control Laboratory at the Massachusetts Institute of Technology. A multichannel system is superior in several respects to individually stabilized amplifiers, in particular, reduced initial cost, smaller size, and less maintenance. The system built at the D.A.C.L. retains the basic advantages of a multichannel system and incorporates several design features that improve its operation and extend its range of usefulness. The features of the D.A.C.L. drift-stabilization system result in better performance than previous multichannel systems and are particularly advantageous in generalized high·accuracy analogue facilities. The computer described in this paper is a large. binary. gt'neral purpose. dif!:ital computer designed jointly by JB:\I and MIT engineers. ,Some of the features which make its operation efficient are a 6.0 microsecond cycle magnetic core memory, an indexing system for automati~ally modifying instruction addresses, an input-output control which eliminates all interruptions of computation including those caused by access time, a dual arithmetic element which opt'rates on two sets of operands simultaneously, a buffer drum system which handles high data rates by writing in the first empty register to pass the heads. and a one-half micro-second-perbit multiplication. The prototype has been in operation for over a year. Computer Design to Facilitate Linear Programming f.ombined Analog and Digital Computing Techniques for the Solution of Differential Equations R. C. Gunderson, mathematician, Remington Rand The growing importance of linear programming in business. industry, and go\ernment has presentt'd the users and manufacturers of high speed digital computers with an exciting facet of computer application. However, the question im· mediately arises as to how we might better deiign future computers to exploit the possibilities of this powerful tool. The actual needs are few and mathematically simple. Furth· ermore. these needs are extremely compatible with those logical properties desired by logicians. However. consideration must he gi\en to the a(hantages gained timewise by more efficient use of high spt'ed storage Illt'dia. Secondly. the occurrence of many zeros in the linear systems involved suggests the use of some form of zero suppression, conserving both time and storage space. These are but a few of the requirements which should be gh en consideration in the building of the future machines of our industry. Hurney~ supervisor, Electronic Development, Massachusetts Institute of Technology One of the difficulties in the use of an electronic analogue computer for the solution of ordinary differential equat ions invoh-ing variable coefficients is its relative inability to perform certain multiplications rapidly and accurately. In in",tances where a variablt' mm~t bt' multiplied by a function of another variable, this difficulty is particularly apparent. This paper describes how a digitally stored table of functions may be used with an analogue computer to solve this general class of ordinary differential equations. 12:30 - 2:00 p.ln. Chairman John F. Ai~craft An experimental interpretive routine for the JB:\I 705 has been prepared which monitors the instructions of the code in question and gives a complete history of the computer action as a result of this code ..-\ll instructions. or only selected instructions. depending on a console switch setting. will be monitored. The history will be written directly to a line printer or to magnetic tape for later printing. again depend· ing on the position of a console switch. Digitizing pulse duration modulated data at a rapid rate and presenting it in a suitable form for data reduction has been a problem for some time. The Magnavox Series 200 Converter is designed to accept PDM data recorded on magnetic tape, automatically digitize it, and record the digital information on the magnetic tape in a form suitable for input to a digital computer or other data reduction equipment. This paper covers the design and operation of the Series 200 converter. The machine has been in operation for a period of eight months and this experience will be reported. Paul A. IB~l 705 CONFERENCE LUNCHEON Venetian Room' Haanstra~ senior project engineer, International Business \Iachines Corp. Speaker' Edward Teller, professor of physics, lTniversity of California. Berkeley 2 :00 p.lll. - 5 :00 p.nl. DESIGN, PROGRAl\Il\IING AND CODING Terrace Room Chairman Paul L. Morton, professor and chairman, Division of - 28 - Considerations in ,Uaking a Data Gathering System Computer Compatible Bill L. Waddell, G. ::\1. Giannini &: Company. Inc. The paper discusses the design problems facing the Data Systems engineers required to produce a Data Collection System that will be able to enter easily a computer. Some empirical formulas are presented with a discussion of how to use tJIt'''e formula ... The many recording tools and their application to Data Systems preparing for entry into com· puter" are dt'''crilwd with a discussion of the place of each recordin~ clt'\ ict'. The second section of the paper is a critical analysis of Four Data Recording or Gathering Systems dt',ignt'd to go directly to a Digital Computer. Computers and Automation 2 :00 p.m.-5 :00 p.m. SCIENTIFIC APPLICATION Nob Hill Theater Chairman n. H. Lehmer, professor, Department of Mathematics. The proper role of the small machine in handling scientific and engineering computations will be illustrated by examples of the performance of the Burroughs EIOl in the aeronautical field. The basic parameters of the machine, speed, capacity, etc., will be outlined in order to provide a framework to evaluate the specific applications which will be presented. University of California, Berkeley Using a Variable Word Length Computer Jor Scientific Calculation Traffic Simulator with a Digital Computer s. y . Wong, Fred Gruenherger, numerical analyst, General Electri(' Company E. H. Coughram, I.B.M. branch manager, Richland. Washington Variable word length alphameric machines are designed primarily for commercial data processing. One would expect relatively low efficiency and some compromise with desired operations in using such a machine for numerical analysis work. Just the reverse seems to be the case. The variable word length affords many outstanding advantages which far outweigh two small disadvantages. Integrated Data Processing with the Univac File Computer R. P. Daly, data handling systems engineer, Remington Rand The manner in which the Univac File Computer meets the needs of input/output flexibility, balanced internal storage of large capacity, and adequate external storage for integrated data processing is discussed with examples. Unusual Problems and Their Solutions by Digital Computer Techniques Lawrence Rosenfeld, head, Operations Research, Mathematical Services Group, Melpar, Inc. A number of case histories of "unusual problems" will be outlined and discussed. These will include: (1) The determination of stock trends for the speculative department of a large brokerage house; (2) Baseball forecasting and its use by a gambling house; (3) The determination of optimal trucking routes through a given traffic congestion pattern. A Fixed-Program Data-Processer Jor Banking 0 perations Jack Goldberg, research engineer, Computer Laboratory, Stanford Research Institute The internal programming differs from that of the typical stored-program computer in that the sequence of operations in the main is pre·programmed by the wiring of the machine rather than by coded instruction words. The machine is intended for an "On-line" operation reqbiiing a large variety of simultaneously occurring processes, with severe requirements of accuracy and reliability. A Progress Report on Computer Applications in Computer Design S. R. Cray, electrical engineer, Remington Rand R. N. Kisch, electrical engineer, Remington Rand 9:00 a.m. - Noon This paper summarizes the logical properties of a set of magnetic switch building blocks and presents mechanized pro· cedures for computer design using these elements_ Symbols are defined for representing the building blocks in algebraic equations. All of the combinatorial ground rules for proper electrical operation are reduced to equation formats. A threephase design program is then outlined which has been used for processing those portions of computing systems which are constructed of these building blocks. CIRCUITS Nob Hill Theater Chairman J. D. Noe, assistant director, Division of Engineering -1 Topological Application oj Digital Computing Machines Ascher Opler, Research Department, Dow Chemical Company A method has been devised by which networks can be reo duced to a digital code which represents the topological configuration. Logical programs may be written which en· able a digital computer to analyze these coded representations for the existence of specified sub-networks. The code appears capable of extension to other elementary operations in applied topology. Research, Stanford Research Institute A One-Microsecond Adder, Using One-Megacycle Circuitry A. Weinberger, National Bureau of Standards J. L. Smith, National Bureau of Standards An analysis of the functional representation of the carry digits in the addition process shows that the one-megacycle circuitry of SEAC and DYSEAC can be organized logically to permit the formation of many successive carries simultaneously. A parallel adder utilizing this principle is developed which is capable of adding two 53-bit numbers in one micro· second, with relatively few additional components over those required in a parallel adder of more conventional design. The Transfluxor: A Magnetic Gate with Stored Variable Setting Thursday, February 9 9 :00 a.m. - Noon Philco Corp. This paper presents a method of traffic simulation with a digital computer as means to (1) study traffic control systems (2) plan new roadways and (3) supply information for theoretical studies. APPLICATIONS Terrace Room Chairman Roger Sisson, Canning, Sisson & Associates Applications oj the Small Digital Computer in the Aeronautical Industry Huhert M. Livingston, sales engineer, Burroughs Corp. Edgar L. Lyons, sales engineer, Burroughs Corp. - 29 - Jan A. Rajchman and Arthur W. Lo, RCA Laboratories, Princeton, New Jersey The transfluxor is a new magnetic gate with stored variable setting. It comprises a core of magnetic material with a nearly rectangular hysteresis loop and having two or more apertures. The control of the transfer of flux between the three or more legs of the magnetic core provides novel means to store and gate electrical signals. The characteristics of a representative two-aperture transfluxor are described. Computers and Automation Bilateral Magnetic Selection Systems for LargeScale Computers Amir H. Sepahban, section engineer, Government & Characteristics of the 'RCA Bi~mac Computer A. D. Beard, L. S. Bensky, D. L. Nettleton, G. E. Poorte, Radio Corporation of America Industrial Division, Philco Corporation In the RCA Bizmac Computer the input and output data are stored on magnetic tape stations. Five input and ten output trunks are available. The Computer has a high·speed memory of 4,096 characters, and an auxiliary memory of 32,000 characters. The latter serves as the main instruction storage. The machine has an instruction complement of twenty· two different instruction types, many of which can be subject to minor variations at the discretion of the pro· grammer. Instructions are of the three-address type. Selective writing of information on a chosen channel of a large memory system (e.g., a magnetic drum memory) and selective reading of information from one out of many such memory channels can be accomplished by use of a single two-way magnetic pyramid made solely of high quality magnetic saturable cores_ A description is given of a working magnetic selection unit used in a large inventory control system with a few thousand magnetic drum channels. T he Megacycle F erractor Programming the Variable-Item-Length RCA Bizmac Computer T. H. Bonn, department head, Component Research & L. S. Bensky, T. M. Hurewitz, A. S. Kranzley, R. A. C. Lane, Radio Corporation of America Development, Remington Rand The design characteristics of the computer were arrived at after a careful study of the processing needs of business problems. The most important of these characteristics is the fully variable length of data items on magnetic tape. The manner in which the computer handles variability in all of its aspects has provided a uniquely adaptable tool for commercial applications. The writing of programs for the Bizmac Computer is therefore also unique in many ways. The ferractor is a magnetic amplifier designed to replace vacuum tubes in digital computer pulse circuits. Operation at information rates as high as 2% megacycles with moderate power gains and power levels has been achieved. Ferractors are readily adaptable to modular construction. Using them as building blocks, the control and arithmetic sections of computers can be economically constructed with a minimum number of circuit types. 2 :00 p.m. - 5 :00 p.m. RCA BIZMAC SYSTEM Terrace Room * ___________ Chairman E. S. Calhoun, manager, Electronic Data Processing Re- ICc - - - - - - - - - - • SPE.CIAL ISSUES OF ('('C'OMPUTERS AND AUTOMATION" search, Stanford Research Institute Purpose and Application of the RCA Bizmac System Tbe June issue of "Computers and Automation" commencing with June, 1955, is aspecia1 issue, "Tbe Computer Directory." W. K. Halstead, J. W. Leas, J. N. Marshall, E. E. Minett, Radio Corporation of America The RCA Bizmac System has been designed specifically to meet the data-processing needs of large business operations. Therefore, a number of novel features were incorporated into the RCA Bizmac System. These include an entirely new concept of data-recording variability, a much higher level of system integration than heretofore provided, and several special-purpose machines designed to take much of the burden of the major computer, which .is part of the system also. For details about the next com put e r directory, see "The Computer Directory, 1956: .Notice. ~, Functional Organization of Data in the RCA Bizmac System A. D. Beard, W. K. Halstead, J. F. Page, Radio Corporation of America The characteristics of business data will be discusse.d, as well as the influence which these characteristics had on the basic planning of the RCA Bizmac System. It was desirable to provide complete variability of data-item and message length in storing information on magnetic tape. Th!s led to compression factors as high as five in the storage of certain files when compared with the earlier fixed-word, fixed,block concept. The System Central Concept in the RCA Bizmac System J. A. Brustman, P. T. O'Neil, J. L. Owings, Radio Corporation of America The purpose of the System Central is to integrate the elements of a Bizmac System and to provide for controlling the performance of each element so that the combination functions in proper concert. In the Bizmac System, specialized supervi'lory equipments have been designed to apply over-all control and direction. 30 - .. How Commercial Controls FleKDwriters® and Auxiliary Equipment are used for MANUSCRIPTS INSTRUMENTATION Articles. We desire to publish articles that are factual, useful, understandable, andin~sting to many kinds of people engaged in one part or another of the field of computers and automation. In this audience are many people who h ave expert knowledge of some part of too field, but wOO are laymen in other parts of it. Consequently al~iter should seek to explain his subjec t, and show its context and significance. He should define unfamiliar terms, or use them in a way that make s their meaning unmistakable. He should identify unfamiliar persons with a few \\I'ords. He shruld. use examples, details, comparisons, analogies, etc., whenever they may help readers to understand a difficult point. He should give data supporting his argument and evidence for his assertions. We look particularly for articles that explore ideas in the field of computers and automation, and their applications and impl~ca~ions. An article may c~rtainly be contr~ versial if the subject is discussed reason ably Ordinarily, the length should be 1000 to 4000 words. A suggestion for an article should be submitted to us before too much work is done. FLEXOWRITER FEATURES Prints at 100 words per minute Remote Non-Print Control Automatic Feed Back pulses Automatic Timing pulses Printing up to 280 characters per line Control voltage 90 or 48 VDC Remote Color Shift Control Automatic Tab and Carriage Return Programmed Format Control Transmit or Receive directly Available in 5, 6,7, 8-Channel Tape APPLICATIONS Computers-Input Output Recording and Logging Systems Machine Tool Controls Automatic Calculations Conveyor Controls Data Reduction Systems Punched Tape Verifying Data Preparation Punched Tape Conversion Punched Card Preparation Process Control Systems Technical Papers. Many of the foregoing ~quUe ments for articles do not necessarily apply to technical papers. Undefined technical terms, unfamiliar assumptions, mathematics, circuit diagrams, etc., may be entirely appropriate. Topics interesting probably to only a few people are acceptable. Fiction. We desire to print or reprint fiction which explores scientific ideas and possibilities about computing machinery, robots, cybernetics, automation, etc., and their implica~ ~9Qt1nued on pags! CONTROl Commercial Controls punched paper tape equipment is now used in offices, factories, and a wide variety of research and development projects. The Flexowriter automatic writing machine will print, punch and read paper tape. In addition, it will transmit or receive information directly. Many types of equipment are now using the Flexowriter for direct data input and output-to prepare program tapes for input-to capture output data in printed form. The Auxiliary Motorized Tape Punch, when cableconnected to other equipment, records data in punched paper tape. The Auxiliary Motorized Tape Reader reads punched tape to direct the automatic operation of other equipment. We are interested in articles, papers,reference information, science fiction, and discussion relating to computers and automation. To be considered for any particular issue, the manurer~t should be in our hands by the fifth of the preceding month. Reference Information. We desire to print or reprint ref~rence information: lists, roste~ abstracts, bibliographies, etc., of use to computer people. We are interested in, m a king arrangements for systematic publication from time to time of such information, with other people besides our own staff. Anyone who would like to take the responsibility for a type of reference information should write us. ~nd AUXILIARY MOTORIZED TAPE READER .• V · ~~ ~ ......................................... WRITE for complete Information. - I . Please mention the application in which you are interested. COMMERCIAL CONTROLS CORPORATION 1 Leighton Avenue • Rochester 2, New York Dept. CA·56 Sales and Service offices in principal cities listed in classified telephone directory under "Typewriters-Automatic" ti' - 31 - LR.E. (continued from page 19) ~omputers and Automation netic one-word registers of the recirculating type. A six-bit static register stores instruction digits and drives a switching matrix to produce needed control signals. A t1Yo-phase square-tvave clock operating at 660 kc drives the two standard types of series magnetic amplifiers. These amplifiers. and all other circuitry in the computer are packaged on plug-in, printed~viring panels of seventeen types. Memory is provided by a magnetic drum rotating at 16,500 rpm Ivhich stores 2,000 machine 1fords. 39.4. Sponsored by the Professional Group on Electronic Computers. To be published in Part 4 of the IRE Convention Record. Electronic Computers III-Symposium on the Impact of Computers on Science and Society Chairman: Theodore H. Bonn, Sperry Rand Corp., Philadelphia, Pa. The Univac Magnetic Computer -- Part II. Megacycle Magnetic Modules B.K. Smith, Remington Rand Univac Division of Sperry Rand Corp., Philadelphia, Pa. Through intelligent packaging, mass-production economy is possible on even unique or s p ecialized computers. The magnetic amplifier, ad~t able to all normal vacuum-tube computer functions, has proved a satisfactory module. This pap e r .points out that, in consideration of the properties of magnetic materials, a nelY philosophy of 'design is required. This philosophy entails close collaboration of.logicians, designers, packaging engineers. and research physicists, from the beginning of the development of a computer. Through nelf miniaturization techniques and improved magnetic materials, Sperry Rand Corporation has obtained reliable ·results from magnetic amplifiers at frequencies over 2 me. Development of useful computer forms to replace common logical circuits is discussed. 'The recent development of digital and analog computers has had a profound effect on science a and technology. Science has been given a new tool -- the ability to perform calculations that were heretofore considered impossibly complex and time consuming. In addition, the development of computers as a branch of teChnology has contributed to the generation of new ideas, which in turn are affecting other disciplines. How are these events shaping the course of scientific research and technological development? On what new goals are scientists focusing their attention nOlY that computers are available to them? HOIV Itlll these new tools of science affect our daily lives? What problems will they present and what benefits does the future hold? A panel of distinguished speakers will talk on the above problems. At the conclusion of prepared talk~, there will be a round table discussion of the problems raised. 42.1. A.V. Astin, National Bureau of Standards, Washington, D.C. 42.2. R.E. Meagher, University of Illinois, Urbana, Ill. The Univac Magnetic Computer -Part III. Drum Memory 42.3. D. Sayre, International Business Machines Corp., New York, N.Y. V.J. Porter, S.E. Smith, and M. Naiman, Remington Rand Univac Division of Sperry Rand Corp., Philadelphia, Pa. 42.4. J.W. Forrester, M.I.T., Cambridge, Mass. Criteria for the selection of the type of logic are considered, and single-layer logic is offered as' the optimum logic for this application. 39.5. SESSIGN XLII A magnetic drum-memory with a capacity of 110,000 bits at an operation frequency of 658 kc is described. Storage includes 24,000 bits at a maximum access time of 0.9 millisecond and 72,000 bits at 3.6 milliseconds, with the remainder for sprocket and timing functions. SESSION XLVIII Sponsored by the Professional Group on Instrumentation. To be published in Part 5 of the IRE Convention Record. The memory is sealed in helium to protect it against corrosion, reduce input pot~er, and improve heat dissipation. The drum's high speed (16,600 rp~ and high pulse-density ensure the short access times and the high bitrate. A method of magnetic-head construction is described tvhich makes for a compac t struc ture and facilitates the precise locations of heads in respect to the drum. Instrumentation II 48.3. Extending the Versatility of a Laboratory Magnetic Tape Data-Storage Device . A.V. Gangnes, Ampex Corp_, Redwood City, Calif. The fields of research and development en- . compass a vast number of processes requiring many different methods of acquiring, storing, and evaluating test inffrrmation. This paper des-'32 - (continued on page 44) • • ENSINEERS . . . .1 pioneers in INERTIAL NAVIGATION Immediate openings for Supervisory and Staff positions as well as for Senior Engineers. Engineers, and Associate Engineers, experienced in: SYSTEMS EVALUATION GYROSCOPICS DIGITAL COMPUTERS ACCELEROMETERS TELEMETRY GUIDANCE SYSTEMS STABILIZING DEVICES SERVOMECHANISMS AUTOMATIC CONTROLS THERMODYNAMICS OPTICS ENVIRONMENTAL RESEARCH TRANSFORMERS RELIABILITY WEIGHT CONTROL • ARMA, recognized for its accomplishments in the fields of navigation and fire control, is a leader in the development of Inertial Navigation. This new system deals solely with space, time and acceleration ••. acting independently of external influences. Creative engineering of the highest order is required to develop components making Inertial Navigation possible: accelerometers to measure acceleration; integrators to convert this information into velocity and distance; D'IIrOs to provide directional reference and hold the system stable; computers to calculate course-to-steer and distance-to-go. Components must meet rigid weight and size requirements ... and function with undreamed-of accuracy. ARMA, one of America's largest producers of ultra-precise equipment, offers unlimited opportunity for engineers to help in this great endeavor. Challenging projects and ARMA's extensive supplementary benefits make an ARMA career doubly attractive. ARMA engineers are currently working a 48 hour week at premium rates to meet ao critical demand in the Defense Dept's missile program. Moving allowances arranged. Send resume in confidence to: Technical Personnel Dep't. 2-500 Division of American Bosch Arma Corporation Roosevelt Field, Garden City, Long Island, N. Y• RCA offers opportunities IN MISSILE TEST Data Reduction for ~ MATHEMATICIANS ~ STATISTICIANS ~ PHYSICISTS ~ ASTRO-PHYSICISTS e. RADIO CORPORATION OF - 33 - Degree plus experience in reduction of test data, applied mathematics, statistical techniques, or observatory practices. Positions now available on Florida's central east coast. Liberal company benefits-Relocation assistance. For information and arrangements for personal interview, send complete resume to: Mr. D. E. Pinholster Employment Manager, Dept. N-14E Missile Test Proiecf RCA Service Co., Inc. P.O. Box 1226 Melbo"urne, Florida AME~ICA TORONTO COMPUTER Computers antI Automation {continued from page 25} series expansion, -Aurmning the terms until they becorne less than 10 • Let (COl) =1; (002) :=: 10, -8 and let x b e available in ZOI. Then the following seq u en c e computes ~ -" : 001 002 003 004 005 006 007 008 009 OVER OVER ZERO MULT ADDN DIVD ADDN KOMP TRNS COl.O COl.O Z04.0 Z02.0 COl.O Z02.0 Z03.0 Z02.0 004.0 000.00 000.00 000.0 ZOl.O Z04.0 Z04.0 Z02.0 C02.0 000.0 Z02.0 Z03.0 000.0 Z02.0 Z04.0 Z02.0 Z03.0 Z05.0 Z05.0 BSET WRTE It was seen how iterations on numbers in the elec tronic s tore are done td th the help of the LOOP instruction. In such iterations the B counters are set and reduced automatically by the program. When it is desired to perform a sequence of iterations on DRUM positions this automatic controlof the B counters is impossible and it is necess a ry to set and reduce them by explicit instructions. For example the instruction 000.5 011.0 The basic instructions transfering da ta between the X or Y page and DRUM storage locations have been described. The following two examples illustrate the use of B instructions when modifylng drum positions: Wri te the X page on DRUM pos i tion {001..- (B4) where B4 is to contain the integer n: The answer is available in Z03. BSET temporarily, the content of a B line until required. When needed, the B-line may be restored with a BSET instruction. 000.5 000.0 XOl.O puts (X0l) in B line 5, the second address being a dummy. Specifically it is the content 0 f the exponent line only of XOI that is involved here. Any B line 2 to 6 inclusive, may be set by either of these two instructions and any store position (X, Y, Z or C) not B-modified, may be used in the third address of the second of these instructions. To reduce the counters, the instruction NEGB 000.5 001.0 001 002 003 004 subtracts 1 from (BS) and NEGB 000.5 000.0 X01.0 subtracts the (exponent line of XOl) from (BS). If it is desired to add a number intoaBline INCB 000.5 001.0 Finally, it is sometimes convenient to store the contents of a B line. This may be done with JOTB 000.5 000.0 2.24 WRTE I NCB TRNS 000.0 000.3 000.4 001.0 000.3 000.0 000.0 xOO.O 000.0 000.0 Miscellaneous Instructions There remain to be described 8 number of instructions which perform miscellaneous operations. 007.2 006.0 X01.0 causes the machine to output seven numbers, e ac h with a six-digit mantissa, two numbers per paper line, from consecutive storage locations, begnming tvi th (XOl). The output may be "print" only, "ll.llldl" , only (for later printing on a separate Teleprinter) or "print and punch", according to the setting of a 3-t'ITay std tch on the console. Apart fro m the specifications in the PRNT instruction, the form& of the printing is fixed, consisting of: 000.0 adds the integer 1 into (BS). 000.4 016.0 001.0 000.4 003.0 BSET LOOP In the last example, note that B3 is used to control the loop and B4 to control the progressNe modification of the magnetic DRUM address. Not e also that the run of DRUM positions lies entirely in the lOtfer half of the range. Some further notes on DRUM selection are given in Section 3.62. PRNT 000.0 000.0 XOO.O If n is to take on the values 1, 2, ••• 15, 16 in succession, these instructions should be embedded into a loop as follows: 005 BSET OOn.O 000.4 There is one important restriction, namely, that the unmodified DRUM location and the modified DRUM location must both lie in the same half of the range. The tt"o halves of the range are numbere d 001, 002, ••• 032 and 033, 034 ••• 064 respectively. 000.0 puts the integer 11 in B line 5, the tidrd address being a durmny. 000.4 001.0 for each number the mantissa (one digit before the decimal point and rounded-off) the mantissa's sign 1 space the exponent modulo 100 (two digits) the exponent's sign 2 spaces X01.0 Thus, in the above example, if t'tfhich stores (BS) in the exponent line 0 f XOI. Possible addresses for the last four instructions are as for the B-set instructions. If more than five B lines are simultaneously required in a program, JOTB may be used to store, - 34 - (XOD (X02) (X03) (X04) = = -= = (X05) :: ... 0.01234567 - 1.234567T 123.4567 - 12345.67 -t 1234567. (continued on page 3(.) PHYSICIST or ENGINEER ANALOG COMPUTER' HEAD We're looking for an individualist. We'd like him to be of a high order of analytical ability, and to have already become proficient in computer programing in support of activities in the fields of reactor physics, heat transfer, hydraulics, and mechanics. After that he will take the reins; and using his administrative abilities, show us that this is one position that is definitely fitted to ~he man. If you have the vision to keep on looking ahead and are realistic enough to keep a staff looking ahead right up there with you, then we invite you to write in confidence to General Electric. For now, more than ever, people like you are creating the positions they will continue to grow with. Personal Interviews will be arranged with all selected candidates. In writing; please include your experience, age. academic background and technical references. Mr. E. P. Galbraith Technical Personnel Placement GENERALe ElECTRIC Richland, Washington - 35 - " ) 1DR~1D Cml.PUfER Computers and Automation (continued from page 34) (X06) '::: (X07) = - 123456789. + 12345678912. the output will be printed as: 1.23457.. 1.23457+ 1.23457+ 1.23457+ 0202 ... 06+ 10+ 1.23457- OO~ 1.23457- 04+ 1.23457- 08+ instructions as it finds them by a corresponding sequence of instructions in real machine code. When the last pseudo-instruction has been read in there must be some way of indicating to the machine that it is now to make final preparations for obeying the program.* This is accomplished by the instruction QUIT An extr~ line feed terminates the PRNT instruction. A£ symbol and one other character associated l'11ith checking procedures, will terminate each printed line. All these should be ignored. See Section 3.65. Note that the last line need not contain a full complement of numbers. (8tm>N ~ 64 where m is the number of digits per mantissa and N is the number of numbers per paper line. Some further facilities of the PRNT instruction are given in Sections 3.5, 3.66, and 3.67. The instruction VOID 000.0 000.0 000.0 The instruction 000.0 000.0 inserts a stopU into the program, and is generally used to terminate a piece of calculation. In real FERUT code there are two stop instructions designated as /G and /L. The console has two switches corresponding to these instructions. If a /G (or /L) stop is encountered, and the /G (or" /L) switch is on, the machine t-vill come to a halt until instructed to go on by means of the console switch marked MANUAL PREPULSE. In normal running procedure, the /G is on, the /L off. It is also possible to program a dynamic stop by meam of a short loop of instructions from which control can emerge only by manual intervention. For calculations which proceed over a considerable time there is some danger that information on the electronic store will be lost. This danger can be lessened by inserting the instruction 000."0 COKE inserts a dummy instruction into the program, and is obeyed without action. Of the four letters used to designate the function of an instruction only the first is actually interpreted by the Transcode program. tErefore an unwanted instruction may be over~vritten on the tape by converting its first character to a V, but it should be noted that for sequencing purposes there is still an instruction present. In particular the tape representations for P and V are as shotvn in Fig. 2.1, so that a P may be l' 000.0 which must be the last instruction in any set. Note that QUIT is never used for any other purpose than this~ HALT In general, from 1 to 21 numbers (inclusive) may be output by one PRNT instruction, the number of digi ts per mantissa may be betlveen 1 and 12 (inclusive), and the address of the first number may be any X, Y or Z position, not B-modified. The decimal point is always printed, even if only one digit it called for. The length of a paper line must be adjusted so that the total number of digits, signs, spaces etc. does not exceed 64. Since the format is fixed, this means that 000.0 Fig. 2.1 V changed to a V by inserting one additional hole. This makes it very convenient to use the PRNT instruction for printing out intermediate results during the stage of program development, and then to convert PRNT to VOID once the program is known to be correct. As already explained, Transcode works b y having its program read tape and replacing pseudr - 36 - 000.0 000.0 000.0 at intervals corresponding to a calculating time of 5 to 10 minutes, causing certain constants, instructions etc. to be refreshed automatically from their more permanent form on the drum. The times for individual operations are listed in Appendix III. Note that all three addresses in each of the four instructions VOID, QUIT, HALT and COKE are dummies. They need not necessarily be punched as zeros, provided that they look like permissible Transcode addresses. This allolvs any Transcode instruction to be converted into a VOID, as mentioned above. The last instruction, FNTN, is a particularly important one, allowing simple utilization of Transcode library programs which have been written to carry out standard mathematical processes. Suppose, for example, that is necessary to evaluate the sine of an angle during a Transcode program. There is available a standard routine for this, and it may be assigned an arbitrary function number, say 001, for the problem in which it is to be used. During input, in the manner explained in Section 2.3, the library tape for TC:SIN will be read into the storage location reserved for the first function. The" FNTN instruction must also specify, in addition to the function number, the addresses where the argument is to be found (continued on page 44) ROBOT SHOW STOPPERS Did you see our story Magazine, March 19, pp Bryant magnetic in L if e l73--T76? drums From time to time you may nee d t 0 help organize a display in a business show including'some device that you hope will "STOP" every pers 0 n attending the show and make him notice your display - a device which may be called a "SHOW-STOPPER". In addition to publishing the magazine "COMPurERS AND AUTOMATION", we have for six years been developing and constructing "ROBOT SHOW-STO PPERS", small robot machines tha t respond to their environmen tan d Two of them behave b.y themselves. are: RELAY MOE: A machine that will play the game Ti t-Tat-Toe with a human being, and ei ther win or draw all the time, or (depending 0 nth e setting of a st'lli tch) will so metimes lose, so as to make the game more interesting for the hum an being (tvas at the I.R.E. Show, in Guardian Electric's exhibit; see picture in Life Magazine); SQUEE: An electronic robot squirrel that will hunt for a "nut" indicated by a person in the audience, pick it up in his "hands", take the nut to his "nest", there leave it and then. hunt for more nuts (see picture in Life Magazine); . Besides these we have other small robots finished or under development. These machines may be rented for shows under certain conditions; a~ so, modifications of the small robots to fit a particular purp 0 se are often possible. To: Berkeley Enterprises Inc., 513 Ave. of the Americas, R156 New York 11, N. Y. Please send us more informa ti 0 n about your ROOOT SHOW STOPPERS. The advertising appliclltion tile h,w.:: i n for semi·permanent storage of data in digital computers or for use as delay lines • Designed to purchaser's requirements • Drum runout ,00010" T.LR. or less • Air bearings or super'precision ball bearings • Belt drive or integral motor drive • Speeds to 100,000 RPM • Capacities to 5,000,000 bits or more • Vertical or horizontal housing • Head mounting surfaces to suit • High density magnetic oxide or electroplated magnetic alloy coating mind is: _________________________ Complete Information On Request-write: From: (Organization) (Address) (Filled in by: Name, Title, Date) BRYANT GAGE and SPINDLE DIVISION P. O. Box 620-K, Springfield, Vermont, U. S. A. DIVISION - 37 - OF BRYANT CHUCKING GRINDER CO. MEeHANI ZED MUSE (continued from page 13) oline! George! Computers and Automation , THE WEARY NOMAD FRAMES HIS ROAD APART REPELLED BY CAUTION ON THE SHATTERED BRINK TO HER HE FLINGS HIS INCANDESCENT HEART UNCERTAIN OF THE SNOW HE LONGS TO SINK Rodney!" shouted the Doctor. Doors flew open on all sides of the room, and four agitated young people hurried to the Doctor's side. "Who has left the cards in alphabetic a 1 order?" thundered the Doctor, his mild b 1 u e eyes fairly flashing. "I ask you please t 0 look --" "Note the appearance of an word," the Doctor pointed out. "Ve put the vocabularies in alphabetical order every few weeks for checking and replacing worn cards," explained the Doctor, moppillJ his br01'1. ''My staff has the strictest orders to randomize them before returning them to the files. It ees routine. What has happened?" The machine shuddered slightly and paused, as if gathering strength for the final assault. AS ANXIOUS CANDLES FORGE A FLEETING CLUSTER THE NUBILE MOllIS REVOLVE IN PRAISE OF BUSTER "These must be the checking cards," said one of the girls -- El iz abeth, I think -- in a faint voice. "They must have been put in the file by mistake. I'll get you the right ones immediately -- I put them through the Rando~ izer myself. I'll have them here in two minutes. tI "A very fair example," said the Doctor, tearing off the sheet and handing it to Willy. "A little polishing, perhaps -- but you have some fine lines there." While we were t'laiting for the correc t cards, I asked Dr. Yaffee what the Randomizer was. By the time the Doctor had made himse I f clear, Elizabeth returned, announcing that the checking cards had been removed from the card feed and the correct cards inserted. 0nc e more, Willy pushed the "ON" std tch, and w e gathered breathlessly about the machine~ the type bars started to move. mE HOUR IS TWICE A CAT ON VELVET ROSE WHO MELTS THE MOON UNTIL llIE WILLOW SINiS UNFOUND DELIGHT STANDS MIERE 1HE LANTERN GROWS AND MEETS mE GLASSY SHORE ON DOWNCAST WINGS "Holy -- jumping Willy. MO$es," 0 b s e r ve d b Ii gat 0 r y FORGOTTEN FEARS CREEP DOWN THE BROKEN WALL DIM SHADOWS 1WIST THE CONTOURS OF mE SEA THE WIND REPEATS llIE EARNEST SEAGULLS CALL AND GIVES llIE DREAMING RAIN THE MORNINGS KEY Four shocked young faces bent 0 v e r the printed lines, and four trembling voices repeated them. "The Randomizer, al though only a by-product of the Poem-Writer, ees a most valu a bl e adjunct." he said proudly. "It was built at a cost of half a million dollars, and has since proved useful in many other fields. Briefly, it accomplishes in a few seconds t\fhat it tvould take you many hours to do if you sat down in the middle of this room and tossed the card s about repeatedly. Furthermore, it put s th e cards in random random order, rather than i n ordered t;andom order. I make myself clear?" 0 As I left to catch my commuters' trai n, Willy was murmuring the lines over to himself, a puzzled look in his eyes, but p rid e 0 f authorship, I feel sure, dawning in the depths of his nubile incandescent heart. - END - * - - - - - - - - - )'~----------* EDITOR'S NOTES (continued from page 4) 1. Copy should be typed in columns four inches wide td th single spacing using a bl ack ribbon on white paper. It is very desirable to have a carbon paper ribbon on the typewr.irer since that makes a great deal of difference in tvhat the camera sees, and what the p r i n ted pages look like. It is also desirable of oourse that the typewriter have well-aligned keys. 2. We do not require exact right-hand justification of lines of'typing: one or at most two characters over, and one, two, three, or four characters less are all satisfactory to us. 3. Any untypable information, such as mathematical formulas, should be entered td th good handwriting using jet black ink, and mak(continued on page 42) - 38 - ,'/,' f ' . /' ' ~, ~ /~/~ '; '/: NOW! dependable relays for printed circuits Maybe you, too, have been awaiting availability of a gpod relay for direct insertion into printed circuits. Now Automatic Electric can solve your problem with a miniature relay that is just right. 120 million operations, without a single readjustment or relubrication! That's what you get from this rugged, improved Series SQD Relay, because it features a special heavy-duty bearing and bearing pin. Also a recess in the bearing plate retains an adequate supply of lubricant for long-term lubrication of the bearing pin. Consider these additional advantages: 1. The sections of the terminals that insert into the printed circuit board are NOT brazed or welded into place, but are integral parts of the coil terminals and contact springs-thus preyenting internalloss in conductivity or continuity. 2. Terminal design permits direct plug-in of the relay into a printed circuit board, ready to be secured in place with any acceptable soldering technique. Usually the desired contact spring combination, or pile-up, is sufficiently large so that additional mounting (support) of the relay is not necessary. SQD Miniature Printed Circuit Relays are available with many different contact spring arrangements, and for a multitude of applications. Springs can be made of phosphor-bronze, ~~Bronco" metal, or other specialpurpose materials, as required. Of course the long life, heavy-duty features of the improved SQD Relay can be had in the conventional / type of plug-in relay, if tegular sockets are preferred for use, whether in printed circuitry or other applications. To get complete details, write: Automatic Electric Sales Corporation, 1033 West Van Buren St., Chicago 7, Illinois. In Canada: Automatic Electric (Canada) Ltd., Toronto. Offices in principal cities. AUTOMATIC ELECTRIC NEW PATENTS Ht\YMOND R. SKOLNICK, Reg. Patent Agent Ford lost. Co., Div. of Sperry ~nd Corp. tong Island City I, New York The following is a compilation 0 f pat e n t s pertaining to computers and associated equipment from the Official Gazette of the Uni te d States Patent Office, dates of issue as indicated. Each entry consists of: patent number / inventor(s) / assignee / invention. January 31.1956: 2,733.004 / John E. Richardson, Pasadena, Calif. / - / An electrical computer for solving the equation El E2=E3 F4 of Ifhich El. E2, and E4 have a predetermined amplitude. 2,733,008 / John B. D'Andrea and Herbert M. Heuver, Dayton, Ohio / - / A digi tal c 0 nverter. 2,733,383 / George C. Wilson, Chatham, N. J. / - / An electrical time delay apparatus. 2,733,388/ Adolf W. Rechten and Bria n M. Bellman, 'Taplol\f, Eng. / British Telecommunications Research, Ltd., Taplow, Eng. / A magnetic amplifier for effecting the momentary operation of an electromagnetic rela y in response to an impulsive signal. 2,733,391 / Robert H. Mayer, Middle R i v e r, Md. / The Glenn L. Martin Co. ,Middle River, Md. / An integrator. 2,733,392 / Harold M. Wright, Troy, Ohio / - / An apparatus for synchronizing a slave r 0tating element with a constant speed master rotating element at a predetermined angular displacement relative to the master element. 2,733,409 / Saul Kuchinsky, Philadelphia, Pa./ Burroughs Corp., Detroi t, Mich. / A puIs e code modulation system. 2,733,410/ William M. Goodall, Oakhurst, N. J. / Bell Telephone Lab. Inc., Nm'JYork, N. Y. / A pulse code modulation coder. 2,733,425 / Frederick Calland Williams, Timperley, Eng., and Gordon E. Thomas, P 0 r t Talbot, Wales / National Research De ve 10 pment Corp., London, Eng. / A servo contro 1 means for data storage device. 2,733,432 / Jack Breckman, Long Branch, N.J./ U.S.A. / A circuit for encoding a s i 9 n a 1 ampli tude as a code group of digits ign a 1 s in cyclic binary code. February 7. 1956: 2,733,631 / Dan McLachlan, Jr., Salt Lake City, Utah / Research Corp., Nelv York, N. Y. / An optical analog computer using projected light patterns. 2,733,862 / lIans P. Luhn, Armonk, N. Y. / International Business Machines Corp., New York, N. Y. / A binary decode counter. 2,734,160 / Clifford V. Franks, Cleve 1 and , Ohio and Walter J. Brown, Titusville, Fla./ Walter J. Br01.l'n / An electrical system having a controllable converter supply in g power to a load. - 40 - 2,734,162 / Gordon C. Blanke, Sierre Madre, Calif. / Beckman Instruments Inc., Sou t h Pasadena, Calif. / A rectifying and voltage regulating circuit producing a voltage-controlled direct current output from a source of alternating current. 2,734,165 / Carroll W. Lufcy, Silver S pri n g and Albert E. Schmid, Jr., Greenbelt, Md., U.S.A. / A magnetic amplifier with half-wave phase reversal type output. 2,734,168 / Robert A. Zachary and Joh n G. Schermerhorn, Syracuse, N. Y. / Ge ne r a 1 Electric Co., N. Y. / A phase detector circuit responsive to the phase difference between two alternating vol tages of the sam e frequency. 2,734,182 / Jan A. Rajchman, Princeton, N.J./ Radio Corp. of America, Del. / A magnet i c matrix and computing device. 2,734,186 / Frederick C. Williams, Timperley, Eng. / National Research Development Corp., London, Eng. / A magnetic recording system for an electronic binary digital computing machine operating with serial mode p u Is e train signals. 2,734,187 / Jan A. Rajchman, Princeton, N.J. / Radio Corp. of America, Del. / A system for selectively driving to a \.desired magne tic condi tion one or more of n plurality of driven magnetic elements individually identifiable as corresponding to the elements of a matrix arranged in rows and columns. February 14. 1956: 2,734,684 / Harold D.Ross, Jr., and Clarence E. Frizzell, Poughkeepsie, N. Y. / International Business Mac h i n e s Corp., New York, N. Y. / An e 1 e c t ron i c counter made up of cascade connec te d bistable elements. 2,734,692 / Leland P. Robinson, Pas a de n a , Calif. / Electro Data Corp., Pasadena,Cauf./ A tape storage device for data keeping equipmente 2,734,949 / Clifford E. Berry, Altadena, Calif. / Consolidated Engineering Corp., Pasadena, Calif. / A device for automatic a 11 y and periodically correcting the zero d r if t in an amplifier. 2,734,954 / Marshall C. Kidd, Haddon Heights, N. J. / Radio Corp. of America, Del. / A card swi tching device in a data s tor age device. 2,735,005 / Floyd G. Steele, Manhattan Beach, Calif. / Northrop Aircraft, Inc., Ha.vthorne, Calif. / A two-way binary counting circuit. 2,735,021 / Ole K. Nilssen, Collings1vo 0 d , N. J. / Radio Corp. of America, Del. / A magnetic binary device. 2,735,066 / John L. Coil, Richmond, and Rich- AIRBORNE ELECTRONICS AND WEAPON CONTROL SYSTEMS At Ramo-Wooldridge today there exists a wide range of projects intended to aid aircraft in navi~ating to the vicinity of targets, finding the targets, destroying them, and returning safely to base. Work is under way in such fields as infrared and microwave detection, information display, communication and navigation, and analog and digital computing. Some projects are in the laboratory development stage, some in the flight test stage, some in pilot production. Positions are available for scientists and engineers in these fields of current activity: Communications Systems Digital Computers and Control Systems Airborne Electronic and Control Systems Electronic Instrumentation and Test Equipment Guided Missile Research and Development Automation and Data Processing Basic Electronic and Aeronautical Research Good progress is being made in the establishment of facilities and operational patterns that are well tailored to the unique requirements of advanced electronic systems work. The Ramo-Wooldridge Corporation 15730 ARBOR VITAE STREET. LOS ANGELES 415. CALIFORNIA Computers and AutomRtion ard Gundelfinger, San Pablo, Calif. FBerkeley Scientific Corp., Richmond, Calif. IA counting-rate meter. 2,735,082 I Jacob Goldberg, Bonnar Cox, and James E. Heywood, Palo Alto, Calif. I - I A data sorting system. February 21, 1956: 2,735,302 I Arnold T. Nordsieck, La Jolla, Calif. I - I A mechanic a 1 integrating device. 2,735,615 I Harvey O. Hoadley, Rochester, N.Y.I Eas tman Kodak Co., Roches ter , N. Y. I An electronic analog multiplier circuit. 2,735,616 I Harvey O. Hoadley, Rochester, N.Y.I Eastman Kodak Co., Rochester, N. Y. I An electronic multiplier circuit. 2,735,971 I Robert S. Raven, Catonsville, and Harry C. Moses, Baltimore, Md. I U.S.A. IA two speed control circuit for a servo system. 2,735,977 I William M. Webster, Jr., Princeton, N. J. I U.S.A. I An inverter circuit. 2,735,987 I James B. Camp, Fairfield and Coleman H. Watson, Birmingham, Ala. I U.S. Steel Corp., N. J. I A magnetic memory device. 2,736,007 I David E. Kenyon, Cold Spring Harbor, N. Y. I Sperry Rand Corp., Del. I A teledata system for conveying a plurality of intelligence signals to a remote point. 2,736,019 I Clyde E. Vogeley, Jr., and Theodore Miller, Pittsburgh, Pa. I U.S.A. I A phas ecomparator tracking system. 2,736,021 I David E. Sunstein, Bala-Cynwyd, Pa.1 Philco Corp., Philadelphia, Pa. I A s ig na 1 integrating system. February 28, 1956: 2,736,490 I John W. Schneirer, Upper Darby, Pa. I - I A computing de vic e for determining a ratio between two quantitative measurements. 2,136,770 I Joseph T. McNaney, San Diego, Calif./ General Dynamics Corp., Del. I A pr in t e r capable of responding to a source of in put information comprising code and synchronizing signals. 2,736,801,1 Clyde E. Wiegand, Oakland, and ~~en Chamberlain, Berkeley, Calif. / U.S.A. I A distributed pulse height discriminator. 2,736,802 I Lawrence Cranberg, Los Alamos, New Mex. I U.S.A. I A pulse height analyzer system. 2,736,851 I Jean R. H. Dutilh, Paris, Fr./ - I An electromechanical phase-shifter in an angular position data transmission device. 2,736,852 I Eldred C. Nelson, Los Ang e 1 e s , Calif. I - I An automatic digital motor control system for machine tools. 2,736,880 I Jay W. Forrester, Wellesley,Mass./ Research Corp., New York, N. Y. I A multicoordinate digital information storage device. 2,736,881 I Andrew Donald Booth, Fenny Compton, Eng. I The- Bri tish Tabulating Machine Co., Ltd., London, Eng. I A data storage dev ice with magneto-strictive read-out. 2,736,883 I Leonard Boddy, Ann Arbor, Mich. / King-Seeley Corp., Ann Arbor, Mich. I An in- 42 - tegrating reI ay and signal mechanism. I Harold R. Kaiser, Woodlands Hills, Claude A. Lane, Culver City, and Wilford S. Shockency, Torrence, Calif. I Hughes Aircraft Co., Del. I A high speed electronic digitalto-analogue converter system. 2,736,889 *---------___ - FNl --_______ - t:: EDITOR'S NOTES (continued from page ~R) ing sure that there are no faintly or partially written lines. 4. Any guide lines or other notations which are not to be photographed should be written lightly with a blue~riting mechanical pencil, because the camera regularly sees blue as white, and so does not photograph it. 5. If a small mistake is made, it can be erased, and typed over. If a large mistake is made, the correction can be typed on a separate piece of paper and cemented tv! th rubber cement accurately over the mistake. W hat counts is what the camera sees. Rubber cenent should be used not glue since rubber cement does not buckle the paper. 6. For removing untfanted black marks (sucl1 as a blot), a good grade of artist's poster white can be bought at an art store and painted on with a fine small paint brush. Painting white stuff over a mistake is excellent removal of it. 7. Small line drawings may be placed where they occur in the text. Other figures and illustrations should be furnished separately, with no colors except black, gray, and white, and with names like Figure 1, Figure 2, and so forth. The text should contain references to the figures using these designations, so that there is some leetvay as to where the figures may be placed in the printed article. The separately furnished figures when printed may be full size or any size, not necessarily reduced 20%. For in photooffset printing, copy and illustrations can be stretched or shrunk as if they were on rubber sheets. 8. We shall of course inspect master copy that we receive and if necessary make other corrections. But since our style of type will often be different from the author's style of type, it will be a help if the author does as much as possible of his own correcting. Publishing by photooffset is becoming steadily more important. More and more people are preparing their final drafts of anything in a form in lvhich copies can be easily supplied to anybody for any purpose. The day will come I think lvhen the normal t... ay of typing anything will be in such a form that 1 to 5000 copies can be easily made from the first final typing. - El'TI - en PUBLICAT IONS c:: (!Iz: UI UI P 34; LINEA~ PROG~A~MING ~ND COMPUTEnS. Reprint of tHO articles by Chandler Davis, in July and August 1955 "Computers and Automation". A clear, well-written introduction.to linear programming,willi emphash. on the ideas. • .•. $1.20 en c:: (!z:I (!z:I UI UI UI z: C 0 en c:: z: UI en ~ (!Im~ z: z: ::) i= 0 ::) D- D- l- I- := 0 c.J c:: := c.J ~ o c.J UI P 2D: THE COMPUTER DIRECTORY, 1955. 164 pages, 7500 Who's Who entries, 300 Organization entries, and 600 entries of Produc ts and Services for Sale in the Computer Field; 250,000 words of condensed factual information about the computer field, June 1955 issue of "Computers and Automation." ..•• $4.00 p :::c.J c.J UI I- c.J UI ...J UI U z: :z: c.J UI I- co ~ ...J ~ := UI :z: c:: (!I UI UI en c:: UI I- == z: UI t e r S l- ~ := Q UI ::::i DD~ ::) D- := 0 c.J ~ en UI Q ...J ~ c.J Z ~ :z: c.J UI := Design and development groups of Northrop's Computing Center offer additional opportunities in the original development of computing and data reduction- components and systems. Laboratory technicians, electronic engineers and mechanical engineers are needed for the design and development in reconnaissance data systems and computing equipment involving transistors, magnetic decision elements, printed circuits and miniaturization techniques. In fact, you can also: 90893 85202 44393 29081 (Solve for the digits-each letter stands for just one digi t o to 9) . All are new numbles, additions, multiplications, etc.; some easy, some hard. Each wi th two messages, 0 n e open, one hidden. Hints for solution. Good exercises in logical reasoning. •••• $1.00 --------MAIL THIS COUPON---------or copy it A large number of job classifications written specifically for computing personnel provide unlimited opportunities with proper salary and advancement assured. If you qualify for any phase of computer research, design or application, contact: Northrop Aircraft, Inc., 1001 E. Broadway, Hawthorne, California. Phone ORegon 8·9111, Extension 1893. Edmund C. Berkeley and Associates, 815 Washington St., R159 Newtonville 60, Mass. Please send me publications circled and your announcement of publications: 32 en z: z: ~ :z: := 6c:: en Applied mathematicians and engineers are needed as computing analysts for aSSignment to Northrop's analogue computing facility, and too, for the newly expanded digital electronic computer department which provides unparalleled service in the practical solution of complex engineering problems. I 25 c.J I- ::: If computing and data reduction are new to you but you are a qualified engineer, mathematician or a laboratory technician, contact us and learn how you may establish a career in this vital field. TRY II A V E and T R A I N ESE FUN your WIT S =TWVAS WASE ENTNS WYE = VIF 2D c.J If you are already an experienced computing analyst or engineer, you will find work here to interest you. P 25: NUMBLES NUMBER PUZZLES FOR NIMBLE MINDS. Report. Con t a ins collection of puzzles like: I ...J Would you like to join one of the progressive Computing Centers on the West Coast ... where a broad variety of equipment and activities will be a constant challenge? P 32: SYMBOLIC LOGIC, by LB~IS CARROLL. Reprint of "Symbolic Log i c, Part I, Elementary," 4th editioo, 1897, 240 pages, by Lewis Carroll (C. L. Dodgson). Contains Letlfis Carroll's inimi table and entertaining problems in symbolic logic, his method of solution (now partly out of date), and his sketches of Parts II and III, Hhich he never t'Vrote since he die d in 1898. • ••• $2.50 + T II en z: ~ z: ~ z: UI (!z:I en Ien >...J UI UI 34 I enclose $ in full payment. (Add 10¢ per item to cover c..ost 0 f handling and mailing.) It is understood that if I am not satisfied, I may return any item lUi thin five days after receiving it, and you will at once refund my money. My name and address are attached. NORTHROP AIRCRAfT, INC. PIONEERS IN ALL WEATHER AND PILOTLESS FLIGHT 5-A-4;1-A - 43 - I.R.E. Computers and Automation (continued from page 32) cribes a system philosophy of instrument development that provides a device of sufficient versatility that a large variety of these data-handling applications can be fulfilled by a single machine. Attention is given to various encoding methods available, variation in volume of information to be stored, and the need for acquiring anel evaluating analog data on different time bases. Particular emphasis is placed on efficient use of equipment space, component environment, and standardization of components, inasmuch as these factors directly affect reliability, cost, and service abili ty. - END- f(moNTo COMPLTER (cont inued from page ~6) and the result stored. FNTN 001.0 ----------------* Thus the instruction XOl.O YOl.O lvill place sin (XOl) in YOlo In general the second address can be any X, Y, Z or C address, B-modified except for the C case, and the third address can be any X, Y or Z address, B-modified. Function numbers may normally range from 001, 002, •• 015, but a larger number may be had by special arrangement lvi th the operator. Chapter 5 of this manual includes a list of Transcode functions available in the library, with notes on their use. Functions of more than one argument can be accommodated, and even operations such as matrix inversion can be brought into the scheme. (See library descriptions). The FNTN instruction can also be used to divide a long program into convenient segments. Thus, a programmer may construct his Oll1n FNTN's, as described below. In fact, whenever the total number of instructions in one set appreciably exceeds 100, the progra~ should be segmented. * During the tape read-in, Transcode addresses are replaced by machine code addresses. However, this cannot be done for control transfers to instructions yet to come and therefore a directory list of machine addresses is built up during the read-in. Part of the final preparations consists of using this list to fill in the actual addresses where necessary. In this connection it should be . noted that provision is made for at most 64 forward control transfers, but this is not likely to be a restriction in any Transcode program. '# ///G in machine code. (continued on page 45) and which at the same time is a good story. ~ dinarily, the length should be 1000 to 400h\Urds. Discussion. We desire to print in "Forum" brief discussions, arguments, announcements, news, letters, descriptions of remarkable new developments, etc., anything likely to be of substantial interest to computer people. ·~ayments. - END - NOTICE The June 1956 iss ue of "Computers and Automation" will be the second issue of "The Computer Directory". Last year I'Ve published the first issue, 164 pages. Our present plans for the June 1956 directory follow: Part. 1 of the directory in 1956 will be a cumulative "Roster of Organizations in the Computer Field" based on the last cumulative roster (published December 1955, containing about 330 entries) and brought up to date. Entries in this roster will be free. If you knOl4J 0 f any changes, addi tions, or corrections 1" hi c h should be made in the entries, please tell u s. Part 2 of the directory will be the secund edi tion of "The Computing Machinery Fie 1 d : Products and Services for Sale." 0 v e r 600 entries on 21 pages appeared in the firs t edition in June 1955; a considerable increase is anticipated. The previous entries, and blank forms, were sent in February, to suppliers for revie1'/, checking, and additions. A nom ina 1 charge of $6.00 an entry is requested from each supplier in order to help defray the cos t 0 f preparing and printing the directory; but if the charge is not paid, the entry may s till appear in condensed form, if desirable to make the listing complete. Part 3 of the directory will be the third edi tion of the Who's Who in the Computer Field. In the June 1955 issue, about 7500 en t r i e s appeared on 96 pages; of these about 2600were full entries, and the remainder were brie f entries. Our present plans are to p ubI ish only nety or revised Who's Who informati 0 n in the June 1956 directory. Blank forms for nelV or revised entries l'{ere sent in March to a I I computer people l'11e knotv of. A nominal charge of $2.00 an entry or other support of the Who's Who was requested from each person t"hose entry is printed, in order to help defray the cos t of preparing and printing the Who's Who. The main reason for the nominal charges mentioned above is that lve look on the directory as a service to many people in the c 0 mputer field; yet so far it has not paid for itself; and we need to make a com pro m i s e, publishing at least some information ab 0 u t everything that sh9uld appear in the directory, but fuller information for those tv h 0 h a v e shared directly in the cost. *--~~--~~~~--~~--------------------- ¥anuscript N~tice (continued from page 31) In many cases, we make small token payments for articles, papers, and fiction, if the author wishes to be paid. The rate is ordinarily ~¢ a word, the maximum is $20, and both depend on length in words, whether printed before, whether article or paper, etc. THE COMPUTER DIRECTORY, 1956: - 44 - TORONfO COMPUfER Computers and Automation (continued frorr page 44) 2.3 Tape Controls punched on a separate tape from the instructions. In addition to the instructions, there is available in the Transcode system a set of TAPE CONTROLS which facilitates the inputting of instructions and data and organization of the program. When the computer is started by reans of certain settings on the console hand switches, it proceeds to read tape until it encounters one of these TAPE CONTROLS, tthereupon it takes the appropriate action. Blank tape has no effect and may precede any tape control. If any character other than a TAPE CONTROL or blank tape is encountered, a dynamic stop* tvill occur and tape input t"lill cease. The follotdng is a list of the TAPE CONTROLS and a description of their effects. INST ooj is for reading the program. It must be followed directly by j operational instructions of the type described above, the last one necessarily being QUIT 000.0 000.0 000.0. Instructions are . read into the store consecutively in the same sequence as they are punched on the tape, the first instruction becoming 001, the second 002 etc. When it is required to make reference to any instruction the instruction number is specified and it is not necessary to know where the instruction is stored with reference to the "real" machine. ~ote that the instruction numbers and the decimal separator for the B-line are not punched. CNST pqq pqq pqq reads in the set of constants, the first being CO~ the second C02, etc. Assimilation by the program is automatic and takes place during translation. Note the termination by'" to indicate that the last constant has been read in. There can be at most 21 constants per program. NUMB pqq ••••• ~n~ pqq •••••EMARK REG. U. S. PAT. OFF. NATIONAL CASH REGISTER COMPANY 3348 West BI Segundo Blva., Hawthorne, Calif. ELECTRONICS DIVISION - 47 - ADVERTISING IN ~~ICO,MPUTERS AND AUTO,MATION" Memorandum from Berkeley Enterpr1ses, Inc. Publisher of COMPUTERS AND AlJTO\1ATION 513 Ave. of the Americas, New York 11, N.Y. 1. What is "COMPUTERS AND AUTOMATION"? It is a monthly magazine containing articles, papers, and reference information related to computing machinery, robots, automatic control, cybernetics, automation, etc. One important piece of reference information published is the "Ros t e r of Organization~ in the Field of Computers and Automation". The bas ic subscription rat e is $5.50 a year in the Uni ted States. Sin g 1 e copies are $1.25, except June, 1955, "The Computer Directory" (164 pages, $4.00). For the titles of articles and papers in recent issues of the magazine, see the "Back Copies" page in this issue. 2. What is the circulation? The circulation includes 2000 subscribers (as of Feb.lO): over 300 purchasers of individual back copies; and an estimated 2500 nonsubscribing readers. The logical readers of COMPUTERS AND AUTOMATION are people concerned with the field of computers and automation. These include a great number of people tvho will make recommendations to their organizations about purchasing computing machinery, similar machinery, and components, and whose decisions may involve very substantial figures. The print order for the May issue wa~ 2700 copies. The overrun is largely held for eventual sale as back copies, and in the 'case of several issues the over ru n has been exhausted through such sale. 3. What type of advertising does COMPUTERS The purpose of the magazine is to be factual and to the point. For this purpose the kind of advertising Ivan ted is the kind'.that answers questions factually. We recommend for the audience that we reach, that advertising be factual, useful, interesting, understandable, and new from issue to issue. We reserve the right not to accept advertismg that does not meet bur standards. AND AU1~MATION take? 4. What are the specifications and cost of advertising? 'COMPUTERS AND AUTOMATION is pub1 ished on page~ 8~" x 11" (ad size, 7" x 10") and produced by photooffset, except that printed sheet advertising may be inserted and bound in with the magazine in most cases. The closing date for any issue is approximate 1 y th e 10th of the month 'preceding. If poss ible, the company advertising should produce final copy. For photooffset, the copy should be exact I y as desired, actual size, and a~sembled, and may include typing, 'oJri ting, 1 in e dr a tv in g, printing, screened h,'lf tones, and any othe r - 48 - copy that may be put under the photo 0 f f s e t camera without further preparation. Unscreened photographic prints and any other copy requirmg addi tional preparation for photooffset shou I d be furnished separately; it will be prep ared, finished, and charged to the advertiser a t small additional costs. In the case of printed inserts, a sufficient quantity for the issu e should be shipped to our printer, addres s on request. Display advertising is sold in units of a full page (ad size 7" x 10", basic rate, $190) twothirds page (basic rate~ $145), and half pag e (basic rate, $97); back cover, $370; in sid e front or back cover, $230. Extra. for color red (full pages only and only in certain pasi tions), 35%. Two-page printed insert (one sheet), $320; four-page printed insert (tt'lO sheets), $590. Classified advertising is sold by the tv 0 r d (60 cents a word) with a minimum of 20 words. 5. Who are our advertisers? Our advertisers in recent issues have included the follow in g companies, among others: Aircraft-Marine Products, Inc. American Bosch Corp. Ampex Corp. Armour Research Foundation Arnold Engineering· Co. Automatic Electric Co. Bendix Aviation Corp. Cambridge Thermionic Corp. Epsco, Inc. Ferranti Electric Co. Ferroxcube Corp. of America General Electric Co. Hughes Re~earch, and Development Lab. International Business .Machines Corp. Lockheed Aircraft Corp. Lockheed Missile Systems Logistics Research, Inc. The Glenn L. Martin Co. Monrobot Corp. Norden-Ketay Corp. Northrop Aircraft, Inc. George A. Philbrick Researches, Inc •. Potter Instrument Co. Ramo-Wooldridge Corp. Reeves Instrument Co. Remington Rand, Inc. Republic Aviation Corp. Sprague Electric Co. Sylvania Electric Products, Inc. Mathematical Analyst Keith Kersery loads jet transport flutter problem into one of Lockheed's two 701's. On order: two 704's to. help keep Lockheed in forefront of numerical analysis and production con1rol data processing. With two 701 digital computers already,in operation, Lockheed has ordered two 704's to pennit greater application of numerical analysis to cOP1plex aeronautical.problems now being approached. Scheduled for delivery early next year, the 704's will replace the 701 'so 704's and 701~s speed Lockheed research in numerical analysis Much of the work scheduled or in progress is classified. However, two significant features are significant to career-minded Mathematical Analysts: '1) the wi&~ variety of assignments . created by Lockheed's diversified development program and 2) the advanced nature of the work, which falls largely into unexplored areas of numerical analysis. Career positions for Mathem·~tical Analysts Lockheed's expanding development program in nuclear energy, turbo-prop and"jet transports, radar search planes, extremely highspeed aircraft and other classified projects has created a number of openings for Mathematical Analysts to work on the 704's. Lockheed offers you attractive salaries, generous travel and moving allowances which enable you and your family to move to Southern California at virtually no expense; and an extremely wide range of employe benefits which add approximately 14% to each engineer's salary in the form of insurance, retirement pension, etc. Those interested in advanced work in this field are invited to write E. W. Des Lauriers, Dept. MA-3~-5. LOCKHEED BURBANK AIRCRAFT CORPORATION. CALIFORNIA DIVISION CALIFORNIA ADVERTISING IN.DEX Computing Devices of Canada, Ltd., P. O. Box 508, Ot tawa, Canada / Elec tronic Computers / P age 2 / CA No. 135 Ferroxcube Corp., East Bridge St., Saugerties, N.Y./ Magnetic Core Materials / Page 35 / CA No. 136 General Electric Co., {Hanford Atomic Prod uc ts} , Richland, Washington / Help Wanted / Page 35 / CA No. 137 Lockheed Aircraft Corp., California Div., Burbank, Calif. / Mathematical Analys ts Wanted / P age 49 / CA No. 139 National Cash Register Co., Electronics Div., 3348 W. E. 1 Segundo Blvd., Hawthorne, Cal i f . / Digi tal Computer Engineers / Page 47 / CA No .140 Northrop Aircraft, Inc., Hawthorne, Calif. / Employment Opportunity / Page 43 / CA No. 141 The purpose of COMPUTERS AND AUTOMATION is to be factual, useful, and understandable. For th i s purpose, the kind of advertising we des ire t 0 publish is the kind that answers questions, suc h as: What are your products? What are lOur services? And for each product, What is it called? W h a t does it do? Hot\' well does it lvork? What are its main specifications? Following is the index and a summary of advertisements. Each item contains: Name and address 0 f the advertiser / subject of the advertisemen t / page number where it appears / CA number in cas e of inquiry (see note below). Aircraft Marine Products, Inc., 2100 P axto n St., Harrisburg, Pa. / Taper Technique / Page 51 / CA No. 127 Arma Division, American Bosch Corp., Roo seve 1 t Field, Garden City, L. I., N. Y. / Engineering Opportunities / Page 3 / CA No. 128 Automatic Electric Company, 1033 W. Van Buren St. , Chicago, Ill. / Relays for Printed Circ ui t s I Page 39 / CA No. 129 Berkeley Enterprises, Inc., 513 Ave. of the Americas, New York 11, N. Y. / Robot Show Stoppers, Publications, Geniac Kit / Pages 37, 43, 47 / CA No. 130 Bryant Chucking Grinder Co., P. O. Box 62 O-K, Springfield, Vermont / Magnetic Drums / P age 37 / CA No. 131 Cambridge Thermionic Corp., 430 Concord Ave., Cambridge 38, Mass. / Perma-Torg Units / P age 5 / CA No. 132 Commercial Controls Corp. (Flexowriter), Rochester 2, N. Y. / Use Flexolvriter / Page 31 / CA No. 133 Computers and Automation, 513 Ave. of the Americas, New York 11, N. Y. / Back Copies, Advertising /Pages 46, 48 / CA No. 134 Ramo-Wooldridge Corp., 8820 Bellanca Ave., Los Angeles 45, Calif. / Employment Opportunities / Page 41 / CA No. 142 R.C.A. Service Co., Inc., Missile Test Project, P. O. Box 1226, Melbourne, Fla. / Help Wanted / Page 33 / CA No. 143 Sprague Electric Co., 377 Marshall St., Nor t h Adams, Mass. / Pulse Transformer Kit / P age 52 / CA No. 144 READER'S INQUIRY If you wish more information about any products or services mentioned in one or more 0 f the s e advertisements, you may circle the appro pr i ate . CA Nos. on the Reader's Inquiry Form below and send that form to us (we pay postage; see the instructions). We shall then forward your i nquiries, and you will hear from the advertisers direc t. If you do not tvish to tear the magazine, just drop us a line on a postcard. *--------------------------~--------------------* * READER'S INQUIRY FORM Paste label on envelope:J... Enclose form in envelope: '" r - - ------- -- - -- --..., .. --- - - ------ - - - - -- - -- -- - -- - - - - - - - - - - - --~a:J .::.. .c "A- ""'0 0 ell ...... 01 m m ~ CD 0lIl ... ~ -< ;:II; m m m CD 0lIl ." 0lIl 01 m Z : 0 CO z -f iii m .!" ""'0 a.: ~en :r: :< • ~en Your Address? ............. ' ... . Your Organization? zen :;:0 lis Address?..... ~m ~"'O Your Title? ...... "r- FORM .. . .. ... . ~-< ~r- z» Please send me additional information on the following subiects for which I have circled the CA number: OJ m I 6 11 16 21 r- !l 2 7 12 17 22 3 8 13 18 23 4 9 14 19 24 5 10 15 20 25 26 31 36 41 46 7l 32 37 42 q 28 33 38 e 48 29 34 39 44 49 30 51 52 53 54 55 76 77 78 79 80 101 102 103 104 35 56 57 58 59 60 81 82 83 84 85 106 107 lOB 109 40 61 62 63 64 65 86 87 88 89 90 111 112 113 114 45 66 67 68 69 70 91 92 93 94 95 116 117 118119 50 71 n 73 7A 75 96 97 98 99 100 121 122 123 124 105 110 115 120 125 126 127 128 131 132 133 136 137 138 141 142 Ie 1-16!A7 148 129 134 139 144 U9 Y.lQ ttl 140 145 150 z",-U ~r ~ ::!. (") :I! ~ 3: ... ~ REMARKS: ~;o:;~ JIIIIIIIIIIIII gi ~ _ _ _ _ _ _ _ _ 0_ _ INQUIRY Name (please print) ......................... . ?l .$ ~ CO .,-:- go Z "< : fi READER'S >- ~Z ~m ........ ___ --"" . . . . . . . . :...1_________________________________________ _ - 50 - ask about ~@~~ TO BETTER WIRING More Co~fide~ce . .. - Less [}{pe ~se with A-MP Taper Pins, Taper Bloks, Taper Tips, Taper Tabs and Taper Tab Receptacles offer great freedom of circuit design and insure maximum electrical stability. They cut material cost ... cut labor cost ... and reduce assembly time. Taper Technique is benefiting manufacturers of business machines, aircraft, guided missiles and electronic equipment. You, too, can benefit by the A-MP Taper Technique. Consult your local A-MP salesman or write to Harrisburg. @A-MP 1956 Aircraft-Marine Products Inc. I General Office: Harrisburg, Pa. A-MP of Canada, Ltd., Toronto, Canada. A-MP-Holland N.V., 's-Hertogenbosch, Holland Aircraft-Marine Products (G.B.) Ltd., London, England. Societe A-MP de France, Courbevoie, Seine, France HERE'S THE IDEAL TOOL FOR ENGINEERING DEVELOPMENT OF CIRCUITS USING PULSE TRANSFORMERS Type Sprague's new Type lOOZI Pulse Transformer Kit contains five multiple winding transformers. each chosen for its wide range of practical application. Complete technical data on each of the transformers is included in the instruction card in each kit so that the circuit designer may readily select the required windings to give transformer characteristics best suited for his' applications ... . whether it be push-pull driver. blocking oscillator. pulse gating. pulse amplifier. or impedance matching. The electrical characteristics of the transformers in the kit have been designed so that they may be matched by standard Sprague subininiature hermetically-sealed pulse transformers shown in engineering bulletin 5028. For complete infor-mation on this kit, as well as the extensive line of Sprague pulse transformers. write to the Technical Literature Section. Sprague Electric Company. 377 Marshall Street, North Adams. Massachusetts. Sprague on request will provide you with complete application engineering service for optimum results in the use of pulse transformers. Export ror lhe Americas: Sprague Eleclric Inlernalional Ltd., North Adams, Massachusetts. CABLE: SPREXINT.
Source Exif Data:
File Type : PDF File Type Extension : pdf MIME Type : application/pdf PDF Version : 1.3 Linearized : No XMP Toolkit : Adobe XMP Core 4.2.1-c041 52.342996, 2008/05/07-21:37:19 Create Date : 2015:08:22 08:29:46-08:00 Modify Date : 2015:08:22 07:48:23-07:00 Metadata Date : 2015:08:22 07:48:23-07:00 Producer : Adobe Acrobat 9.0 Paper Capture Plug-in Format : application/pdf Document ID : uuid:e747737f-2fb0-0647-bc0c-722d36e6f869 Instance ID : uuid:118ab0bb-0723-de4d-b5ba-32e471641cf2 Page Layout : SinglePage Page Mode : UseNone Page Count : 52EXIF Metadata provided by EXIF.tools