It DIGITAL COMPUTER OFFICEI F NAVAL RESEARCH -Vol ll 1No 2 • MATIIEMATI Sr$ NCfS PIINp Gordon D Goldstein Editor Jean S Campbell Asst Editor Apr 1959 TABLE OF CONTENTS Page No COMPUTERS AND DATA PROCESSORS NORTH AMERICA 1 1 1 Brookhaven National Laboratory Merlin Computer Upton L I New York 2 Datarnatic Div Honeywell 800 Newton Highlands Massachusetts 3 National Bureau of Standards The NBS Multi-Computer System Washington D C 4 Nortronics Airborne Computers Hawthorne California 2 3 5 Philco Corp Transac S-2000 Philadelphia Pennsylvania 4 COMPUTING CENTERS I Franklin Institute Computing Center Philadelphia Pennsylvania 5 2 Georgia Institute of Technology Rich Electronic Computer Center 6 Atlanta Georgia 3 National Bureau of Standards Computation Laboratory Washington D C 4 New York University AEC Computing and Applied Mathematics Center New York N Y 5 The University of North Carolina Research Computation Center 6 6 Chapel Hill N C 6 6 U S Naval Ordnance Laboratory Mathematics Department White Oak Silver Spring Maryland 7 U S Naval Proving Ground Naval Ordnance Computation Center Dahlgren Virginia 7 7 COMPUTERS AND CENTERS OVERSEAS 1 Elliot Brothers Ltd National-Elliot 802 London England 2 European Organization for Nuclear Research CERN Geneva Switzerland 8 10 3 Ferranti Ltd Process Control Computer London England 10 4 Hitachi Ltd HIPAC-I Tokyo Japan 12 5 LEO Computers Ltd LEO I and II London England 14 6 7 15 16 The University of Liverpool Mathematical Institute Liverpool England University of Mainz Institute for Applied Mathematics Mainz Germany 8 National Physical Laboratory ACE Teddington England 9 N V Electrologica X 1 Amsterdam Holland 10 Siemens Halske AG Siemens 2002 Munich Germany 11 Societe d'Electronique et d'Automatisme Courbevoie France 12 Ultra Electric Ltd Tape Merging Equipment London England 13 The Weizmann Institute of Science Electronic Computer Section Rehovoth Israel 18 19 23 24 25 COMPONENTS 1 A B Dick Co High Speed Printer Chicago Illinois 26 2 Lockheed Missiles and Space Div High Speed Digital Plotter Sunnyvale California 26 3 Packard Bell Computer Corp Systems Components Los Angeles California 27 4 Telemeter Magnetics Inc Series RB Storage Units Los Angeles California 27 MISCELLANEOUS 1 Contributions for Digital Computer Newsletter SInformation Thi doc umont has booni approved Approved farpublIc rol - o a s its 16 by Reproduced by the C L EA R I N G H 0 U S E for Federal Scientific Technical Springfield Va 22151 The Under Secretary of the Navy 220 August 1957 28 NAVEXOS P-645 F COMPUTERS AND DATA PROCESSORS NORTH AMERICA MERLIN COMPUTER - BROOKHAVEN NATIONAL LABORATORY - UPTON L I NEW YORK Merlin is a high-speed digital computer being constructed at Brookhaven National Laboratory It is similar in concept to the Maniac 1Unow 'in operation at Los Alamos and is intended to provide on a limited budget speeds and capacities competitive with presently available computers Initially the electrostatic memory of Merlin will hold 12 288 random access self-checking 48 bit words with an access time of 8 microseconds however provision is being made for addressing over 64 000 words Construction of Merlin began in late 1956 at which time the cost of magnetic cores offset their obvic 3 advantages in terms of speed and reliability Recent developments in the design and manufacture of electrostatic storage tubes indicates their use at a density of 6 000 bits per tube within acceptable tolerances It will operate in fixed or floating points with the number representation N 2 8 yx with 41 bits for x including sign and 5 bits for y including sign The remaining two bits of a number may be used as tags or markers which may be detected either programwise or automatically by the computer In addition to three shifting registers which are essential for arithmetic operations the arithmetic unit of Merlin will contain four non-shifting registers which will be used as fast access temporary storage The basic add time will be 3 microseconds with a shift time of 1 microsecond per stage or 1 microsecond per simultaneous 8 stage shift The computer operates asynchronously and an average floating add time of 7 microseconds and an average multiply time of 90 microseconds are expected A Merlin word will hold one instruction which may contain one or two memory addresses and up to three addresses if the fast access registers are involved The fetching of an instruction is performed during the execution of the previous instruction in order to minimize the access time involved Automatic address modification by means of six index registers has been provided and following the design of MANIAC II a separate register has been provided to hold the contents of the control counter upon transfer It was felt that Merlin as a research tool should have flexible provisions for manual interventions and easy manual access to the various parts of the machine is provided Input to Merlin is achieved via paper tape magnetic tape and a typewriter Output is via paper tape magnetic tape the typewriter and a 600 lines per minute on-line printer Provisions have been made for incorporating up to sixteen magnetic tape units Viewed from an engineering standpoint Merlin is unique As a transitional machine bridging the gap between the Thermlonic and Solid State Eras it utilizes vacuum tubes crystal diodes and transistors The Arithmetic and Control Sections are largely tube operated the memory presents an interesting amalgam of tubes and transistors The approximately 8 000 tubes and transistors along with their associated circuitry are contained in 31 free standing racks distributed over a floor space of 150 square feet Throughout the design emphasis is placed on accessibility and ease of maintenance The input-output equipment power supplies and cooling system occupy an additional 600 square feet HONEYWELL 800 - DATAMATIC DIV - NEWTON HIGHLANDS MASSACHUSETTS The Honeywell 800 is a transistorized medium-scale data processing system which is medium-scale in size and price only Its magnetic tape speeds and internal operating speeds are faster than those of any other medium-scale system For business applications it has the ability to perform dccimal arithmetic For scientific applications it can perform binary arithmetic either fixed-point or floating-point An exclusive feature called Automatic Parallel Processing gives the ability to perform several independent programs simultaneously Thus -1i- 11104 IttIoIII hentobinit qually adapitahlo it ave' ie'Ifli' anid Ixisinvoop appid tat ions it can dto iot h typos of jobs uinnultanomotely Am many Ito litht pro gruema may becproe'asod III parallel No speolal prolrammning Is novessary and Individuawl pri' jtaen miay be' starlod must atopieid hidopend ally The 1100 incorporate conspiet 0'rt'king f rom Input thIsrough output atid inc luode Orthoetrnie Contirol oxvitunivo Ilotonywoll techimquesr for ifconaletaefr reet itsi of loot or' gjarbled A powurful 1hroo-addroak order strue'turo vonnorvo'w time' and snivnsory capacity and fact thtat n progrsamminsig A full cotmplinanstt of autunsatict progIramming and library routines - IIl bo availablo it every user Pitrallet Procesasing is atbroakthrough that exactsa maximum offiviency fromt all uflita of ilia #ystvni It to a Combination of two nghineering achievemlenlts called Traffic Control and Muiti-prIogrami Control Traffic Control In a dws ivo which mionitors all peripheral activities oif tile system including magnotic tapes and the console and provides the' proper channel connections at the' proper time betwoen these devices andi the' central processor An many as eight input and eight output chan nelo nay be activo nimiultaneoutily Multi-program Control coordinates the demands of up to eight completely Independent programsn running in parallel The internod operating speed is 30 000 three-address operations per second This Is equivalent to 40 000 two-address or 60 000 single-address operations per secmVd A series of numoers can be accumulated at a rate of 125 000 per second Core storage is available in modules of 4 096 words and one two three or four modules may be employed Each word of core mtc rage Is individually addressable Parallel transmission Is employed in sending words to and from core stoi age An entire word can be sent to or from the memory in six microseconds The memory is the basic storage unit for both data and instructions High-speed random-access drum storage is available as optional equipment The word consists of 54 bits of which six are used for checking The 48 Information bits anay represent an 11-declimal-digit number with its sign sev eral smaller decimal numbers with signs for each eight alphabetic characters or a combination of these A word may also be Interpreted as a 44-bit binary number with its sign or as an instruction Using the floatingpoint option a wvord may represent a sign bit a seven-bit exponent and a 40-bit miantissa in binary form In the instruction word the information bits are divided logically into iour sections which are interpreted as aji operation code followed by three addresses Each address in an instruction may be designated as absolute or indexed A total of eight Index registers are available to each program The ability to mask words allows most internal processing instructions to work witn fieldR4 of variable length Each program may designate a group of 32 memory locations as masking registers Such a designation may be changed by the programmer at any point in his program Thus an essentially unlimited number of masking registers is at his disposal THE NBS MULTI-COMPUTER SYSTEM - NATIONAL BUREAU OF STANDARDS WASHINGTON D C At the National Bureau of Standards a new digital data processor has been designed as a flexible pilot facility for exploring complex processing information retrieval and control problems of importance to the Government Because the system is intended for research on such a wide range of applications - ranging from automatic searching and interpreting of -2- infor mation rv'cerurt real -thiln uchgeduliing and cont1rol ofticommrcial aircraft traffic - it is ch''trAclrired by a variety of features not ordinarily associated with a singleiinatallation It 1a1 not 01 1y a high coniputatiion rate %nda wide rp rtoire of internal procesning operations but aiso vory flexible alind powerhfl control capabilities ftor conmmunicating with human opera- torm and with devico• external to tho myatni The ovor alli system devised by A L Leiner W A Note J L Smith and A Weinberger of the l ata l'rocessing Symtoems Division In so integrated network of three independently programmed vomp uters each of which i s ipeciaily adaptod for performing certain particular clannso of mathiinimatical or logical operations All throt computers operate concurrently and intercommunicate with vach othor in a way that permits them to work together collaborativuly on the solution of problenms being tackled by the system The primary computer a high-speed general-purpoeo coniput• r carries out most of the high-precision arithmetic calculations and logical processing operations The smaller secondary computer acts as its partner keeping track of various parts of the program and carrying out short-word operation- usually manipu- lati•ns on address numbers or other red-tape information which it supplies automatically am rieded to the primary computer Each of theso computers utilizes only 16 basic instruc tons thus providing a simple code structure However because many variations of the formats are provided a wide variely Of operations can be performud with these few instructions Both computers acting c operatively can carry out special complex sorting or search operations The third computer has charge of information flowing into or out of the system This computer may be supplied with information about the anticipated needs of the primary computer for external data and ciui then search the external storage or input-output devices convert the data in these devices to the proper form for interiial use and feed it into the primary computer at the right instant All three computers as well as all the external units share access privileges to the common high-speed internal memory which is linked to the input-output and external storage units via independent trunks for effecting data-transfers The system can accommodate up to 16 such input-output trunks along which data can flow concurrently though only two trunks are planned for the initial installation of the system Using two such trunks it is possible to maintain two continuous streams of data flowing simultaneously between any two external units and the internal memory without interrupting the data processing program The system can• operate with a wide variety of input output devices both digital and analog either proximate or remotely located The external control capabilities of the system enable it to supervise this wide family of external devices and on an unscheduled basis to interrupt or redirect its over-all program autuinatically in ordur to assist or manage them Such impromptu interchanges of information between the system and the devices or people exLernal to it can occur at the instigation of either the external world or the internal system or both acting jointly Automatic controls regulate all the interplay among the different parts of the system and guard against the possibility of internal traffic jams The system will run more than 100 times faster than SEAC on most types of programs and more than 1 000 times as fast on some programs AIRBORNE COMPUTERS - NORTRONICS - HAWTHORNE CALIFORNIA Nortronics Division of Northrop Corporation presently has in production a fully transistorized airborne digital computer APAC which has been fully flight tested and environmentally iualified It has been operational in both manned and unmanned aircraft since November 1957 and has accumulated some 200 hours of successful airborne operation This device is a medium speed D D A consisting of 375 germanium transistors and 1200 germanium diodes D C power supplies are an integral part of the computer Inn t is from punched pnppr tPe module deriving the required D C voltages from 115 volt 400 cycle primary power source All circuitry is constructed on removable etched circuit boards of which a total of 20 are utilized The memory device used is a non-cokitact drum consisting of 5 channels Tht complete computer weighs 100 pounds its volume is 2 cubic feet and requires 132 watts of 115 volt 400 cycle single phase power -3 - Nortronics also has under dovclopmeint an alr'iorne digital computing system which is designed to perform the omnputation function for the Lightweight Inertial Navigation System LINS This computing system combines the advantage# of integral and incremental computer techniques in an integrated digital system Computations for sensor quantizing navigation gyro control inertial instrument erection and pilot's display of present position range to target and hoading to target are performed within the digital system The complete computing system including input-output equipment weighs 50 7 pounds and occupies a volume of 1 3 cubic feet Designed for ia severe airborne environment the system is presently in the assembly stage and is scheduled for completion during the summer of 1959 The division hits alsu recently completed programming of the IBM 704 computer for a multi-stage decision process The operation which formerly was carried out manually and with a small general purpose computer taking 2-1 2 hours now requires less tha 20 seconds Nortronics is also presently developing a computer for an advanced guidance system application TRANSAC S-2000 - PHILCO CORP - PHILADELPHIA PENNSYLVANIA installation of the first TRANSAC S-2000 Electronic Data Procassing System was made in November 1958 at the Western Development Laboratory Palo Alto California - a Philco su isidiary Application of the TRANSAC Transistorized Automatic Computer at WDL iR primarily for data reduction The second system will be installed in Philco's Corporate Division Philadelphia Pennsylvania This system will be applied to Philco's commercial data processing Although Philco has not made its TRANSAC commitments public the Naval Supply Center Oakland California has announced its intention to install the system Highlights of the TRANSAC S-2000 see Digital Computer Newsletter April 1957 Mode Parallel Core Memory 4096 - 32 768 words 48 bits Magnetic Tape Speed 90 000 alphanumeric characters or 168 000 decimal digits per second Punched Cards 2 000 cards per minute read High Speed Printing 900 lines per rminute Paper Tape 1 000 characters per second read Simultaneous Operations Provision is made for the simultaneous operation of up to nine input output devices including four magnetic tape units Of the total up to five may be communicating with the central computer concurrent with processing A unique feature the Universal Buffer Controller provides common buffer storage and control for off-line data conversions and on-line punched card and printer operations In so doing the Universal Buffer Controller reduces the number of magnetic tape punrned card high speed printer and buffer storage units necessary in a system The Universal Buffer Controller also allows for the simultaneous operations described above There may be four Universal Buffer Controllers on-line and each can accommodate two tape units and up to five punched card and high speed printer units The character transfer rate between the central computer and a buffer controller is 90 000 alphanumeric characters per second -4 - COMPUTING CENTERS COMPUTING CENTER FRANKLIN INSTITUTE PHILADELPHIA PENNSYLVANIA - In January ID59 The Franklin Institute Computing Center started its third year of operation The bhsic equipment-a Univac I System acquired from the Sperry Rand Corporation through a Iease purchase arrangement-is maintained by their own engineering stuff The Mathematicm Analysis Section formerly a division of the Laboratories for Research and Development but now an integrated part of the Computing Center provides analysis and programming services as required In all respectr the second year of operation was highly successful Machine performance and production scheduling exceeded expectation and in the field of technical and systems research progress was most satisfactory Much of the data processing scheduled on the machine during 1958 was handled on a sub-contract basis The Institute itself is a non-profit organi•ation primarily noted for research and development in the mechanical arts and sciences Consequently the majority of service bureau Jobs directly contracted for by the Center were of a mathematical or engineering nature Many involved Institute personnel additionally qualified as experts in fields other than compLting During 1959 the Institute will continue to provide machine time and personnel for back-up facilities research one-shot and repetitive data processing Jobs of whatever nature required by our users They also hope to serve as a means of introduction to computers for those industries or organizations which while interested in electronic data processing have for one reason or another delayed their entry into the field The Institute is more than a little disturbed by the increasingly critical content of many articles dealing with the concept of large or medium scale computing systems currently evident in a number of national periodicals management and consulting publications While such publicity is embarrassing to the industry as a whole its effect is particularly evident among the less experienced or potential users of the type normally contacted by a service organization such as The Franklin Institute Through judicious application of the large and comprehensive library of canned routines and the proven record of reliability established by Univac I it is hoped to couvincingly demonstrate that industry of modest size and endowment can economically convert chosen segments of its operation to electronic means It is the Institute's belief that demonstration of the feasibility of this approach devoid of high initial installation rental and personnel costs is one of the simplest and most effective rebuttals to the skepticism being generated by the currently fashionable practice of singling out one or two poorly managed or unsuccessful installations and concluding through broad sometimes irresponsible generalizations that computers in their present stage of development are little more than expensive playthings for government agencies and a handful of giant corporations One such canned routine successfully utilized during 1958 and to be emphasized throughout 1959 is an economic time series analysis Seasonal Adjustments by Electronic Computer Methods A preliminary seminar for area businessmen revealed much interest and a real need for the type of information available through use of this routine Surprisingly enough several of the firms contracting for this service had computing facilities of their own but had never considered this particular area of application Others new to the use of computers were favorably impressed by the economy and ready availability of the routine Another routine Big Ben - 25 for solving systems of linear algebraic equations up to and including order 25 has been in constant use since its development by the staff In many cases solutions are delivered on the same day the input data are received at the Center With respect to this and many similar routines it may be safely said that the facilities of the Center are no further away from the area engineer or mathematician than the phone at his elbow -5 - RICH ELECTRONIC COMPUTER CENTER - GEORGIA INSTITUTE OF TECHNOLOGY - ATLANTA GEORGIA The Computer Center has added a Burroughs 220 digital computer to its existing facility of a Univac Scientific ERA 1101 and an IBM 650 The system includes 5000 10 decimal digit words of core storage a high speed 1000 character per second photo-electric paper tape reader paper tape punch a Cardatron system of one card collator 089 input and one output unit controlling either a line printer 407 or a card punch 523 This complement of equipment is undergoing tests and will be further tested with four additional magnetic tape reel units to be delivered shortly The 220 will enable the Center to expand its research and educational activities This computer has built-in floating point operations as well as effective data processing abilities The addition of a 4 096 core storage unit to the Univac Scientific is nearing completion This Nill supplement the 16 384 word 24 bits drum memory and provides a computing speed of within 20 percent of the 1103A All design modifications were planned and done by the Center's staff The compiler Fortran is being adapted to the 1101 by the Center and is expected to be available for the 220 as well It is now used on the IBM 650 through Fortransit This will make three different computers available through a common language to a user within certain limitations COMPUTATION LABORATORY - NATIONAL BUREAU OF STANDARDS WASHINGTON D C During January the IBM 704 was equipped with half-word logic This makes the machine identical to that operated by the Weather Bureau The Bureau of Standards machine is to be used as a stand-in for the Weather Bureau machine both for overflow work and to avoid delays in regular weather prediction services in case of temporary machine breakdown Early in May a 32 000-word magnetic core memory will be installed while the 8000-word auxiliary magnetic drum memory will be retained The machine will continue to be available to agencies and contractors of the Federal Government After installation of the larger memory it is expected that the hourly cost of machine time will be $215 AEC COMPUTING AND APPLIED MATHEMATICS CENTER NEW YORK UNIVERSITY - NEW YORK N Y - Through the offices of the U S Atomic Energy Commission a committee has been selected to evaluate requests for the use of computing machines at the AEC Computing and Applied Mathematics Center of N Y U Only proposals coming from non-profit institutions can be considered The P'hysical Sciences and Mathematics are the areas for which computing may be approved subject to the availability of time on the Univac and the IBM 704 At present the Univac is much more readily available than is the 704 Interested parties should write for further information to Eugene Isaacson AEC Computing and Applied Mathematics Center New York University 4 Washington Place New York 3 New York RESEARCH COMPUTATION CENTER - THE UNIVERSITY OF NORTH CAROLINA - CHAPEL HILL N C In May 1959 the Consolidated University of North Carolina will install a new Univac Scientific ERA-1105 Digital Computer in the new Physics and Mathematics Building now being built at Chapel Hill Purchase of this machine was made possible through the support and cooperation of the Sperry-Rand Corporation The Bureau of Census and the National Science Foundation -6- Although the computer itself is located on the campus in Chapel Hill it will be ut students and faculty of the North Carolina State College in Raleigh and the Woman's CtA3L of the University of North Carolina as well Plans are underway to make time available olk a cooperative basis to other nearby institutions both in North Carolina and the Southeastern United States In addition to serving as a teaching and research aid in present university areas it is planned to use the computer as a research tool in such frontier areas as language translation automatic programming automatic numerical analysis business statistical and other data processing linguistics analysis numerical solution of part• Q differential equations mathematical logic and decision processes and many other regions of investigation that have become important only since the advent of the high-speed digital computer New courses in high-speed computation both credit and non-credit for graduates and undergraduates are to be given or are under consideration Special opportunities will be available for graduate students to work as programmer-analysts in the new Computation Center either at research assistants or supported by graduate fellowships Undertia iaUes will have opportunities to serve in part-time jobs as computer operators or as data preparation specialists In February of this year Dr John W Carr III formerly Associate Professor of Mathematics at the University of Michigan and formerly President of the Association for Computing Machinery assumed the post of Director of the Computation Center and Associate Professor of Mathematics at the University A particular effort is being made to make research assistantships available to programmers now working in computer installations who are interested in returning to a University as candidates for the Ph D degree in computer-oriented areas Assistantships also are planned that will allow a student to work as a liaison programmer between his particular University department and the Computation Center The first of a series of Conferences on Frontier Research in Digital Computers is being planned for August 17-28 of this year with lecturers from the United States and overseas scheduled to participate The Univac ERA-1105 computer which is on order has floating-point operations over 12 000 words of magnetic core storage over 32 000 words of airectly operable magnetic drum storage 17 Uniservo magnetic tape units punched card and paper tape input-output and a highspeed printer Complete buffering allows independent operation of the main computer unit and the magnetic tapes Similar machines will be in use at the Bureau of Census various Air Material Command installations and Armour Research Foundation MATHEMATICS DEPARTMENT - U S NAVAL ORDNANCE LABORATORY WHITE OAK - SILVER SPRING MARYLAND On 2 February 1959 an IBM 704 was installed at the Naval Ordnance Laboratory The computer is equipped with a 32 768 word memory eight tape units a card reader a card punch a printer and a cathode ray tube display and recorder Off line equipment includes a printer and a card-to-tape converter Present plans call for the 704 to be operated on a single shift basis At the present time it is being used for scientific computations and data reduction problems NAVAL ORDNANCE COMPUTATION CENTER - U S NAVAL PROVING GROUND - DAHLGREN VIRGINIA The UDT Universal Data Transcriber designed and built at the Naval Proving Ground to enable interchange of NORC input and output with a variety of digital media is now undergoing systems checkout -7 - The high speed Charactron microfilm printer built by Stromberg Carlson for attachment to NORC passed its reliability acceptance test in February du ring which 11 000 frames were recorded with only one error that error being detected by printer check circuits The printer had been undergoing debugging and intermittent use since delivery in April 1958 During 1958 the NORC was staffed 6520 hours of which 5194 hours were scheduled for computing and 4868 hours of good computing time were realized Losses during the scheduled time were caused by Arithmetic and Control Tape system CRT memory Printers mechanical Power supplies 2 1% of scheduled time 1 9 1 2 7 4 6 3% COMPUTERS AND CENTERS OVERSFAS NATIONAL-ELLIOTT 802 - EUjLJOTT BROTHERS LTD - LONDON ENGLAND The order code for the National-Elliott802 see Digital Computer Newsletter January 1959 was evolved as a result both of long experience with various types of computers and of close collaboration between engineers and programmers It is extremely simple yet contains a large variety of orders Careful logical design has er sured that there are no exceptions to the rules The functions are numbered in a logical way that is learned very rapidly with no conscious effort on the part of the programmer Each order refers implicitly to the single accumulator and to one store address The comprehensive order code provides a set of 64 different instructions Orders are of the single address type with two orders per word There is an extra digit in each word which is used for automatic modification of orders B-modification B'y means of this digit any location in the store may be used as a B-line A B-lined order takes no longer to obey than the same order without a B-line The ability to modify their own orders is one of the features that make automatic computers so powerful In previous computers facilities for doing this automatically have been limited to a few registers often less than eight The 1024 B-registers in the 802 give the ser tremendous advantages by simplifying the programming of extremely complex problems Reliability is achieved by careful electronic design and by the use of solid state devices The store consists of a matrix of magnetic cores Cores are also used in conjunction with transistors to form the basic logical elements in the machine By tese means long periods of trouble-free operations are ensured The basic logical elements which are completely standard consist only of a magnetic core a resistor and a transistor with printed-circuit connections These units are themselves mounted on printed-circuit plug-in boards The printed-circuit techniques fur her increase reliability and the system of plug-in units pioneered by Elliott Brothers makes maintenance very simple Only three keys are used to control the operation of the machine This extreme simplicity virtually eliminates the oossibility of error due to incorrect manual operation of the controls A loudspeaker indicates the machine is calculating and produces a sequence of tones according to the Instructions being obeyed A comprehensive library of subroutiies is provided with the machine This minimizes the programming req'-ired for the solution of any special problem The dimensions of the basic computer are little greater than those of a desk a nd it can be used In a normal room Weight is low no special ventilation or temperature control is necessary The 802 can be installed in the most convenient locatioi for ease of access by office laboratory or control room staff 8- The computer is a stored program machine which operates from a single level ve -y fast access store This means that all the inst uctions and data are availatle immediately wherever they are situated in the store Thus minimum access coding is not required and the computer is always operating at its full speed No time is wasted in transferring information from a backing store to the working store since the computer has such a large rapid access memory Construction The computer consists of an assembly of individually-ventilated steel cabinets with an attractive grey hammer finish and of convenient deskheight These cabinets contain the logical units on plug-in boards the core store stabilized power supplies and the built-in control console Lnput By 5-hole punched paper tape at speeds up to 170 characters per second Output By 5-hole punched paper tape punched at approximately 25 characters per second and subsequently interpreted at 10 characters per second Optional Input and Output Punched card reader manual keyboard input analogue and digital recording mechanisms additional tape readers and punch Special devices designed to customers' requirements can be provided Storage Magnetic core store capacity 1020 words of 33 binary digits each plus 4 words of fixed orders Access time negligible ir cluded in operating speeds Arithmetic Unit Digit rate Notation arithmetic mode Word Length Order Code Single address two orders per word B Modification Any location in the store may be used as a B-modifier When the B digit is present the contents of Address 1 are added to the Order 2 before it is obeyed This operation takes no extra time Operating Speeds Addition Subtraction and 46 other orders 612 microseconds Multiplication and Division 21 4 milliseconds Negative Number Representation Complement with respect to 2 Accumulator 33 binary digits There is an Auxiliary Register of 32 binary digiLs used with the Accumulator for double length working Power Requirements 2KVA Dimensions The 802 is L shaped One arm contains the store and arithmetic unit and the other arm comprises the control console and the input and output circuitry The power supplies are housed in a separate free-standing cabinet 166 500 per second Binary Serial 33 binary digits Store and arithmetic ui it Co isole and input output Power unit 7' 4 x 2' 6 22 x 76 2 cm 5' 2 x 2' 9 138 5 x 83 8 crm 2' 9 x 2' 6 83 8 x 76 2 cm A ' uiits Lrc2' 8 81 3 cm high An 802 computer was delivered to Panellit Inc Skokie Illinois luring January 1959 and will become the central processing center for the Panellit 609 dt lugging Pystem -9- EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH CERN GENEVA SWITZERLAND - Computer A standard Ferranti Mercury Computer has been installed in the Scientific and Technicai Services Division of CERN The computer completed its acceptance tests on 13 October 1958 There is core storage for 1024 40-bit words and 16 384 words on magnetic drums One word represents either a floating point binary number 10-bit exponent 30-bit fractional part or two single-address instructions There are seven 10-bit B-registers and a comprehensive set of instructions for manipulating 10-bit quarter-words Times for floating-point arithmetical instructions are Transfers Add or subtract Multiply 120 microseconds I 180 300 Division is by sub-programme and takes 3 8 milliseconds All other instructions including those for 10-bit arithmetic take 60 microseconds Input and output is by 5-level punched tape 200 characters-per-second reading 30 characters-per-second punching A 60 character-persecond Teletype punch is being installed Auxiliary equipment A Dataplotter 1133 B system Electronic Associates has been delivered This plots directly from punched tape Programming The computer is available to all of the scientific staff at CERN and most programmes have been written by the physicists or engineers using the Autocode scheme developed at the University of Manchester England Larger problems are programmed in machine code by specialized staff attached to the computer Basic sub-programmes are permanently available on the magnetic drum and a library of programmes on tape is being built up in collaboration with other Mercury users Staff The computer is at present working one shift a day only with occasional evening shifts The present staff of 11 is made up as follows 6 MaLhematicians and programmers 2 Operators and 3 Maintenance engineers Enlargement of staff to about 20-25 is in progress PROCESS CONTROL COMPUTER - FERRANTI LTD - LONDON ENGLAND The first transistor electronic digital computer designed to provide fully automatic control of many industrial processes is being developed by Ferranti Ltd It is about the size of two sm 11l office filing cabinets and the first production models are expected to be tivailable in 1960 At present in the prototype stage it is technically known as the PCTC Process Control Transistor Computer The transistor logical circuits in the computer have been thoroughly tested and an experimental model has controlled a machine which simulates a plant process for more than 2 000 hours operation without component failure Additional vibration and temperature tests have also been carried out to simulate the most rigorous conditions the computer is likely to meet in industrial use The price of the new computer will probably be in the range of £20 000- £50 000 depending on the size of the installation Input for the computer can be in the form of shaft rotations voltages pressures temperatures and so on which will be converted to digital form for processing b the computer However if a very large number of input channels are required it may be necessary to add a further cabinet to house additional selection and conversion equipment It is not possible to state precisely how many input channels can be handled on account of the variety of possible types of input As a guide however it is quite feasible to think In terms of several hundred inputs of 4 or 5 different types Both electronic and mechanical switching between input channels will be used and the corresponding selectton times will vary between microseconds and milliseconds It should be noted that each input channel can be directly addressed and as a consequence there is no delay as would occur in a sequential scanning process - 10 - I Output will consist of different types of signals and analogue currents which can be used directly for control purposes In addition to the input and output of information in analogue form numbers can be Inserted manually by means of a small keyboard for direct input to the computer and output to an electric typewriter for data logging and other purposes is available A further use of the keyboard is for specifying the starting point of the programmes It may be desirable in certain circumstances to have punched paper tape available as a form of input and output Should this be so there would be no difficulty in providing the facility Special attention has been given to providing a comprehensive order code which is easy to use and the result is in many ways similar to the Pegasus order code Instructions are pegged up in the same form as they are written Each order location on the pegboard consists of a row of six groups of eight holes There are 32 rows on a tray and 16 trays in the basic computer making 512 order locations in all The two left hand groups refer to the function the next to the accumulator and modifier while the last three groups specify the address There are four accumulators designated A B C D Accumulator A can be used for either single or double length arithmetic including multiplication while B C and D are single length accumulators They are mostly used for modifying and counting or for arithmetic and red tape operations the main accumulator at the same time performing a multiplication or division The word length is 10 bits but the computer can work with either 10 or 20 bit numbers depending on the accuracy required The 10 bits including sign give an accuracy of 1 part in 500 and for most control applications this is sufficient in those cases where greater accuracy is required double length working 20 bits including sign can be used The order code has been designed so that both single- and double-length arithmetic can be used with equal ease The computer is simple to programme and the comprehensive order code which for example includes eleven types of jump instructions results in compact and efficient programmes The speed of the computer up to 40 000 operations per second is adequate for all foreseeable control applications It is of course possible to store several programmes in tray form in the computer and the operator can select any of the programmes stored in this way as and when required If a completely different programme is required ready pegged trays can be easily slid into position For some applications it may be possible and consequently more economic to have extra trays pre-pegged in this way than to increase the programme capacity of the computer A Library of Programmes can be stored on special perforated cards which slide into thc programme trays Thus the problem of pegging up any given programme becomes a very simple matter Routines which are in constant use such as those required for input and output purposes can be held on permanently wired trays It is likely that this latter facility will only exist for machines with extended programme space In some forms of progamme storage particularly magnetic devices it is possible by the gain or loss of digits for an instruction to become altered and so wrongly interpreted by the computer If this should happen while a control system is operating the consequences might be serious For this reason pegboard programming has been used with this computer particular attention being paid to ensitring that pegs are not dislodged by vibration During the running of a plant under computer control it may be necessary to examine a certain quantity at some stage in the programme Rather than stopping the computer each time such a quantity is required a facility has been provided to examine the contents of certain parts of the computer at a specified instruction in the programme while the computer is still running This is done by setting the number or address of the instruction on the handswitches and when that instruction is obeyed the contents of all the accumulators the multiply divide register and the instruction itself are displayed on indicators on the indicator panel These indicators will remain set until the next time that the specified instruction is obeyed This - 11 - The Magnetic Drum Store has a total capacity of 1024 long delay lines i e about 1 5 million digits There are four drums each having 256 tracks each track storing the contents of one delay line Each drum has 16 read heads and 16 write heads the 256 tracks being obtained by moving the heads as one unit into one of 16 discrete positions Each drum is 6 75 inches long by 5 inches diameter allowing a linear digit packing of 100 per inch The drum is driven by a hysteresis motor running synchronously at 12 000 rpm and is phase corrected so as to rotate exactly once in five major cycles of the machine Punched card and magnetic tape equipment will be provided for input and output The installation will initially consist of two machines A broadside card reader running at 450 cards per minute and a broadside card punch running at 100 cards per minute In order to read or punch 80 columns of a card the computer has an 80 digit input dynamiciser and an 80 digit output staticiser The ACE has been designed and constructed to allow easy maintenance Extensive marginal checking facilities are provided and the chassis units are designed to give complete accessibility to all components valve connections etc The m•chine is housed In 10 cabinets each having a cooled air circulation system Each cabinet is fitted with a rising door which permits immediate access to all the 24 chassis units contained therein The number of valve envelopes is about 6000 X I - N V ELECTROLOGICA - AMSTERDAM HOLLAND The electronic computer X I see Digital Computer Newsletter July 1957 is a product of the N V Electrologica Amsterdam It has been designed and developed in close collaboration with the Mathematical Centre in Amsterdam which has much experience in building electronic computers It incorporates the most modern developments as for example transistors and magnetic cores Both of these have the merit of giving very long service The X 1 can be used for arithmetical and logical operations for business or for scientific purposes it works at the very high speed of 15 000 additions or subtractions or 2 000 multiplications or divisions per second By the application of transistors instead of thermionic valves power consumption has been reduced to a few hundred watts and special cooling apparatus is no longer needed in view of the small amount of heat produced The dimensions of the basic machine are no greater than those of an ordinary writing-desk The storage consisting of magnetic cores comprises an active and dead part both with the same access timn so that the elaborate procedure of optimum programming is no 'inger reeded The active storage is used for variable data and current programmes The dead storage is especially cu_ table for the storingof fixed programmes such as often occur in business administration its component parts can readily be changed and as their cost is low a large atock of fixed programmes ready for immediate use can bex maintained at moderate cost In consequence the size of the raore expensive active storage can be kept smaller than would otherwise be possible The capacity of the standard equipment 512 words active and 512 word3 dead storage can be expanded if necessary by adding supplementary units to a maximum of 32 768 words in all For the input of datb -'taer punched cards or punched paper tape may be employed Output is b means'of punched cards pun hed tape or typewriter The input capacity when tape is used is 150 characters per second the output being 25 characters per second except that the output capacity of the typewriter is 10 characters per second Punched paper tape for use with the X 1 can of course be obtained an a by-product of suitable office machinery e g typewriters adding-and-listing or bookkeeping machines A parametron once eot oscillating coitinues to oscillate even if the input stops As this makes logical process impossible excitation must be suspended so as to halt oscillation pending the next input Thus in the parametron circuit Ili must constantly be interrupted In the mechanical analogy of the parametron action two modes of oscillation are possible onit in which m Is pushed to the left at the start and that in which it is puahed to the right at the start A parametron starting from a small oscillation Induod by an Input signal alone gradually amplifying it with the help of excitation finally acqufres a sufficiently large resonant current The two states are called the 0 phase and w phase or - and phase respectively If to the - phase lo set the 0 of the binary system and to the phase 1 1-beat counting can be done In order to combine parametrons to form various logical circuits the parametrons must be set oscillating by turns They are excited intermittently as mentioned in the foregoing The signals hnwever should not travel backwards Accordingly the method called the three-boat excitation has been adopted All the parametrons used are divided into three groups and each group is excited by tbirns Moreover since the oscillating output of some parametrons must become the input of others the excitations of adjacent groups must overlap more or less If this method of excitation is used the output of the I-beat excited parametrons will become the input of the U1-beat excited paramotrons and the output of the Il-beat excited parametrona the input of the IlU-beat excited parametrons Repeating the series I II I11 I 11 111 in this way signals of the - or phase will be transferred AJC-J The HIPAC-I is a stored-program type of automatic computer for scientific usec T input is by means of perforated tape After initial input all operations proceed automatically and the results come out on a printer For memory a 1024 word magnetic drum is used The HIPAC-I uses about 4 000 parametrons as the logical element Parametrons operate somewhat slower than vacuum tubes but are far more reliable In the HIPAC-I the slower speed of parametrons is remedied by the use of a parallel system The net computaUon time of the computer proper is 4 milliseconds for addition and subtraction 8 for multiplication and 160 for division For the actual computing time however the access time to the memory has to be taken into account in addition to those values The numerical values treated by the HIPAC-I are confined to those between -1 and 1 and tae machine can handle 37 binary digits that is about 11 decimal digits plus a sign digit at the head which expresses whether a number is positive or negative It works on the fixed decimal point system in which the decimal point is placed next to the sign digit Since it is a general-purpose computer it can perform quite a wide range of computations depending on the program There are 38 types of direct arithmetic and control orders in the instruction code including addition subtraction multiplication and division operations Also it is possible to modify the address in the memory designated by the program with a number set beforehand in a register called the cycle counter or with a number varying from moment to moment so as to transfer to another kind of operation This is quite convenient from the point of view of programming and is a proof of its great usefulness Another feature of the machine is its control circuit and its control code By means of a group of control codes various parts of the computer can be controlled directly Thus there is no need to make an initial order as regards to the read-in of information and the memory can be read Into by direct control The control codes can be used to perform a computation by means of tape control The information on the paper tape is first put onto the magnetic drum through the input register and accumulator in the arithmetic and control units - 13 - The computation starts from the address designated by the start control order ond proceeds in turn The instruction code read out of the memory enters the decoder and control circuits and controls the varioun registers and gates The numerical values read out of the drum unter these registers and the accumulator arsd the machine performs the ordered operation The result first appears in the accdmulator and is then transferred into the magnetic drum again or printed out on the printer depending on the next order There are two cycle counters which can revise the content in the address register designated address can also be revised by the content in the address counter The The magnetic drum has 40 heads and rotates at 3 600 rpm The two heads at the left end determine the number of addresses The position of each address is Identified by counting the number of I ' equidistantly written in along the circumference The remaining 38 heads work In parallel to write in and read out 38 bits One-half of that number 19 bits is taken as a short word For an instruction code a short word is always used The address is given to each memory location starting from 0 and ending with 2 047 When the number of counts of the clock pulse counter and the content in the address register coincide the gates of the flipflop group open and the memory content at the desired address is taken out The HIPAC-I has the 1-1 2 address system That is to one operation order one address is designated In the memory but this can also be revised If there is no revision the operation is performed in the order of the addresses unless otherwise specified Each instruction w d is composed of the following three parts 11 bits are assigned Lo the address of an instruction and 3 bits are used for its modification The remaining 5 bits are assigned to the order code which specifies an operation such as addition subtraction multiplication and division etc Some computations performed during the last six months were Calculate the tension and sag of power transmission lines For a series of long transmission wires supported by steel towers the tension and sag of the suspended wires were computed from data such as the distances between the towers and the differences in height of the bearing points It would have taken a hurman 400 days to do what the HIPAC-I completed in 15 hours Another task was the design of a linear accelerator An equation for the motion of electrons involved in this problem was solved for various initial conditions in a little over 10 hours In the design of nuclear reactors the machine has been used to advantage Computations have been made first for a reactor of a simple shape and then for a spherical reactor and next for a cylindrical reactor The calculations center on even values of partial differential equations The machine is being used for the manipulation of experimental data for the study of the method of solving such equations The computer has also been used for various calculations on the numerical control of machine tools mass spectrographic data and even payrolls Among the established routines for the treatment of general problems for the machine are the routine for simultaneous equations of the first order with five to 15 variables routines for various methods of solving ordinary differential equations and function subroutines for square root calculations and for elementary functions such as trigonometric logarithmic functions etc LEO I AND II - LEO COMPUTERS LTD - LONDON ENGLAND LEO Computers Ltd was formed to exploit the computer designed and built by the staff of J Lyons and Co Ltd well known food manufacturers and caterers Active interest in computers dates from 1947 The original computer group was formed in 1949 and forms the - 14- nucieus of the present management LEO I Is claimed to be the first computer in the world to have gone into regular daily operational use on business clerical jobs Since January 1954 it has worked without a break on a 8-day 2-shift schedule The prototype LEO 11 went Into service in 1957 and last year production models were installed for three British companies all were fully working within 3 weeks of the equipment being dismantled in the factory An additional 3 or 4 production models are scheduled for 1959 including one for the British Ministry of Pensions and National Insurance Besides manufacturing computers for sale or rent the company undertakes a large volume of clerical computing work for other companies on contract including production and stock control invoicing payroll and statistical and group pension fund calculations Users of this service include many large industrial concerns insurance offices borough councils etc Most of the work is undertaken to a close time-table with data arriving in unprepared form one day and results being dispatched the next One of the 1959 production models will be installed In a service bureau in Central London In April May 1959 This computer will be equipped with magnetic tape An experienced programming and consultant staff Is engaged with customers applications jobs being charted and programmed on their behalf in all cases Charges for computing are based on the volume of results produced and not on hours workeS so there is every reason for the comaiputers to be operated to maximum efficiency LEO I1 is a binary machine using long life thermionic valves and diode logic The word length is 19 bits with provision for double words of 39 bits pulse repetition rate is 525 kc except in the mercury delay line store where it is 2 1 mc automatic facilities are built in for conversion to and from sterling or decimal notation by single programme instruction There are thirteen immediate access registers including three index registers It may be fitted with up to 4 independent input and up to 4 independent output channels each with its own buffer Input channels read direct from cards and both paper and magnetic tape and output devices include on-line printers card punches and tape recorders Hollerith Bull and Samastronic printers are fitted to current machines Development of link-up for the Anelex printer is in hand All input and output channels handle blocks of 32 words and operate concurrently with normal computing The machine is thus exceptionally efficient on business problems where heavy loads of data and results are normal In addition up to four magnetic drum auxiliary stores of 16 384 words each can be fitted The reading and writing circuits of thp drum are seaprately buffered so that up to 70 transfers oi a block of 32 words to or from the drums can be made per second with negligible loss of computing speed A similar buffering system is applied to the magnetic tape system which uses British Decca tape decks Half inch tape of 100 inches per second is employed and there is full automatic checking of the information written at the time of writing Out of contact recording is used thus eliminating risk of drop-outs due to dust Up to 8 tape decks in 4 cabinets can be coupled to one tape channel Concurrent reading writing and computing is provided for and two separate tape channels can be fitted A fully transistorized core storage system with et unit capacity of 8192 words will be available in 1960 in place of deay incc Veve-al units of this size can be coupled if desired MATHEMATICAL INSZITVTE - r•E UNIVERSITY OF LIVERPOOL LfVtP1OOL ENGLAND - The Mathematical Laboratory of the Department of Applied Mathematics in Liverpool University has been expanded by the addition of a Computer Laboratory An English Electric DEUCE computer is being installed The machine is a standard Deuce with 64 column punched card input and output and will have in addition punched tape input and output facilities The machine is expected to be operational in June 1959 Mr Andrew Young has been appointed Director of the Laboratory - 15 - StU INSTITUTE FOR APPLIED MATHEMATICS - UNIVERSITY OF MAINZ MAINZ GERMANY The digital computer Z 22 was put in full operation during January 1959 The computer has been operated from the beginning with a formula translating system using a restricted version of ALGOL algorithmic language proposed by the joint ACM-GAMM committee ACE - NATIONAL PHYSICAL LABORATORY - TEDDINGTON ENGLAND The ACE or Automatic Computing Engine is the latest digital computing machine to be built at the National Physical Laboratory for the use of its Mathematics Division This new machine with its increased operating speed its larger storage capacity and its many additional functional facilities is a considerable advance as a computing tool on its predecessors the experimental Pilot Model ACE and its engineered counterpart DEUCE Speed It is not easy to assess the speed of a machine from the times taken by the elementa-y arithmetic and logical operations This is particularly true of any machine with high speed stores of less than about 8 000 words on which many problems will demand the use of the backing-up store The times taken for a few basic computations probably give as a ccurate a picture of the overall speed as can be obtained without a detailed study The zeros of a 16th degree polynomial may be obtained in an average time of 16 secs Because of the long word length 48 digits accurate roots even of very ill-conditioned polynomials of this degree will be obtained Polynomials of degrees up to about 250 may be found without using the drum store though no sharp discontinuity in speed results from using the drum on this problem A set of simultaneous equations of order 30 can be solved in about 5 seconds Sets containing no zero coefficients of orders up to 170 may be solved the time taken varying approximately as the cube of the order but diminishing as n becomes large The solution of Poisson's equation V 2 V 4 rp on a square with 400 mesh points may be obtained in about 75 seconds This time is for the direct solution of the finite difference equations and will give values correct to at least 10 decimals After solving one such problem solutions corresponding to different distributions of p may be obtained in about 15-20 seconds each Machine description ACE is a serial computer in which numbers and instructions have 48 biiinry digits The digit rate is 1 5 million per second and therefore its word time or minor cycle is 32 microseconds The main working store consists of 24 mercury delay lines each containing 32 words circulating in a major cycle of 1024 microseconds Rapid access storage is obtained by using mercury delay lines of one two and four words capacity The backing store is four magnetic drums containing a total of 32 768 words The ACE instruction specifies a three address operation of the form 'A function B to D' where A and B are the store addresses of the operands and D is the store address to which the result of the operation is to be sent Instructions also specify a further store address N from which the next instruction is to be extracted The time needed to execute this operation for single length numbers is 32 microseconds and since each operation may be followed immediately by another a maximum rate of operation of 30 000 per second is possible The instruction word uses 47 of the 48 digits divided into groups as follows - 16 - I Group No of digits W 5 A 6 B 6 F 6 D 6 St 1 N 5 J 5 T 5 Ch 2 Wait Number W The minor cycle with which the operation will start Source A Address of operand A Source B Address of operand B Function F Specifies one of 64 functions selected from the following main function groups 0 Logical 5 Multiply Divide Standardize Input and Output Operations 6 Multiply by small integers 7 Add Sul'ract and Discriminate on Result 1 Shift 2 Add Subtract 3 Clear and Add Subtract 4 Instruction Modify Destination D The address to which result is sent In addition to store addresses there are special destinations such as those for discrimination on the result i e conditional transfer of control Stop ST Allows computer to be stopped on selected instructions Used aainly during program testing Next Instruction Source N Store address from which next instruction will come which may be any of the 24 long or some of the shorter delay lines Auxiliary Used for counting prior to a transfer of control Timing J Timing Number T The minor cycle from which the next instruction is extracted Characteristic Ch Determines the length of the operation which may be - a Single Transfer occurs during minor cycle W only b Double Transfer occupies minor cycles W and W 1 c Transfer occupies minor cycles W W 1 W 2 and W 3 Quadruple d Long Transfer starts at W and continues until T inclusive b c ard d are useful for operations concerned with multiple length numbers The Drum Store the Multiplier and the Divider are independent units These are put into operation by the appropriate instructions and then work independently until they have completed their operation This feature provides a parallelism of operation since a multiplication division drum transfer and a series of ordinary operations could be in progress at the same time The three address operation of the machine allows the multiplier and multiplicand to be selected and the multiplication process to be initiated by a single instruction The product is formed in 14 minor cycles i e 430 microseconds The division process is similarly specified and the quotient rounded or unrounded is formed in approximately 1 5 milliseconds The divider also contains an automatic standardizing process for use in floating point arithmetic operations - 17 - The Magnetic Drum Store has a total capacity of 1024 long delay lines i e about 1 5 million digits There are four drums each having 256 tracks each track storing the contents of one delay line Each drum has 16 read heads and 16 write heads the 256 tracks being obtained by moving the heads as one unit into one of 16 discrete positions Each drum is 6 75 inches long by 5 inches diameter allowing a linear digit packing of 100 per inch The drum is driven by a hysteresis motor running synchronously at 12 000 rpm and is phase corrected so as to rotate exactly once in five major cycles of the machine Punched card and magnetic tape equipment will be provided for input and output The installation will initially consist of two machines A broadside card reader running at 450 cards per minute and a broadside card punch running at 100 cards per minute In order to read or punch 80 columns of a card the computer has an 80 digit input dynamiciser and an 80 digit output staticiser The ACE has been designed and constructed to allow easy maintenance Extensive marginal checking facilities are provided and the chassis units are designed to give complete accessibility to all components valve connections etc The machine is housed in 10 cabinets each having a cooled air circulation system Each cabinet is fitted with a rising door which permits immediate access to all the 24 chassis units contained therein The number of valve envelopes is about 6000 X 1 - N V ELECTROLOGICA - AMSTERDAM HOLLAND The electronic computer X 1 see Digital Computer Newsletter July 1957 is a product of the N V Electrologica Amsterdam It has been designed and developed in close collaboration with the Mathematical Centre in Amsterdam which has much experience in building electronic computers It incorporates the most modern developments as for example transistors and magnetic cores Both of these have the merit of giving very long service The X I can be used for arithmetical and logical operations for business or for scientific purposes it works at the very high speed of 15 000 additions or subtractions or 2 000 multiplications or divisions per second By the application of transistors instead of thermionic valves power consumption hzi been reduced to few hundred watts and special -ooling apparatus is no longer needed in view of the small amount of heat produced The dimensions of the basic machine are no greater than those of an ordinary writing-desk The storage consisting of magnetic cores comprises an active and dead part both with the same access time so that the elaborate procedure of optimum programming is no longer needed The active storage is used for variable data and current programmes The dead storage is especially suitable for the storingof fixed programmes such as often occur in business administration its component parts can readily be changed and as their cost is low a large stock of fixed programmes ready for immediate use can bc maintained at moderate cost In consequence the size of the more expensive active storage can be kept smaller than would otherwise be possible The capacity of the standard equipment 512 words active and 512 words dead storage can be expanded if necessary by adding supplementary units to a maximum of 32 768 words in all For the input of data either punched cards or punched paper tape may be employed Output is bj means'of punched cards punched tape or typewriter The input capacity when tape is used is 153 characters per iecond the output being 25 characters per second except that the output capacity of the typewriter is 10 characters per second Punched paper tape for use with the X 1 can of course be obtained as a by-product of suitable office machinery e g typewriters adding-and-listing or bookkeeping machines 18 - For input in the form of punched cards either one or two reproducers may be coupled with the X 1 with a capacity of 7 000 cards per hour for each reproducer A fast reading unit with an input capacity of 42 000 cards per hour may also be used in addition to the reproducers if so desired If all three are used the input capacity is 56 000 cards per hour against an output capacity of about 14 000 cards per hour The fast readear mentioned could also act as a sorter controlled by the X 1 computer In linking an X 1 to a reader or reproducer buffer storage is employed and the computer can continue calculations during card reading and card punching cycles Extensive facilities for operating and controlling the machine as well as for testing programmes are provided with the X 1 Notwithstanding the high degree of reliability of the X 1 both sections of the storage are provided with built-in parity controls The reading and punching apparatus is automatically checked and control can also be applied to typewriter output The price of the X 1 briing6 the use of an electronic computer within the reach of mediumsized enterpri ses The capacity of the standard equipment can be increased by enlarging the storage capacity and by supplementary input and output devices It is possible for example to begin with the use of a siall basic equipment and to expand it gradually as automation proceeds or as the volume of business grows The machine is available either for outright purchase or on hire Instructions Single address code Number System Whatever the input and output medium used punched cards punched tape or tyjpewriter data are fed in or extracted in decimal form the machine automatically transferring to and from its Internal number system The internal use of a binary system makes maximum use of the storage capacity available The machine works with a fixed decimal point Word length 27 binary digits bits or 26 binary digits and one sign digit This gives a storage capacity for numbers to about 67 million Larger numbers up to about 4 5 x 1015 can be stored by making use of two addresses The programming of computations with such large numbers is simple Registers Inter alia two registers A and S each of 27 bits and an address modification register b of 16 bits All these registers are available for adding and subtracting For multiplication and division purposes the A and S registers are used as a double-length register Speed Addition subtraction 64 microseconds multiplication division 500 microseconds SIEMENS 2002 - SIEMENS HALSKE AG - MUNICH GERMANY The Siemens 2002 is a medium-scale transistorized computer It is a general purpose decimal machine with a word length of 12 decimals plus sign and an average speed of 2000 operations per second Special features of the 2002 include three index registers the use of the instruction location counter for address modifications the automatic address substitution and fixed and floating point operations The 2002 has a magnetic core memory of variable size units of 1000 2500 5000 and 10 000 words up to ten memory units can be connected with the central processing unit and a magnetic drum memory with a capacity of 10 000 words Input and outi dAta are handled by punched paper tape punched cards and magnetic tape Magnetic core bffit- s for the input and output units allow the execution of input and output operations simultaneously with the operations in the central processing unit A cathode ray tube unit permits the analogue display of output data Word structure The 2002 is a decimal machine where a decimal digit is represented by a four-g- it--fiiiary number excess-three-code A word can be interpreted by the machine in four different ways namely - 19 - I 1 As an instruction S M 1 R 2 0 3 0 4 0 5 S 6 A 7 A 8 A 9 A 10 A 11 I 12 Decimal 1 can be used together with the sign to mark an instruction Decimal 2 serves several purposes It indicates for instance whether the result of an arithmetical or shift operation is to be rounded or not The operation to be executed is identified by the three decimals 3 to 5 Both decimal 6 address substitution and 12 index tag are used for address modifications The address part of the instructions is given by decimals 7 to 11 2 As a fixed-point number with the decimal point being assumed on the left of the most significant digit the numbers being represented by sign and magnitude 3 As a floating-point number where the mantissa occupies ten decimals 1 to 10 and the characteristic two places decimals 11 to 12 4 As an alpha-numerical expression with two decimal digits characterizing one alphanumerical character Address modification The usefulness of an instruction code depends greatly on the possibility of performing automatically address modifications The 2002 allows for modifying the address part of an instruction in two different ways namely by address substitution and by index register modification this being dependent on the contents of position 6 substitution and position 12 index register modification of the instruction word These two types of modification can be combined and are carried out as follows The control unit of the 2002 include three index registers numbered 1 2 and 3 When executing indexable instructions the number in position 12 of the instruction in the instruction register determines the index register the contents of which is to be added to the address part x position 7 to 11 of the instruction Number 4 in the index tag indicates that the contents of the instruction location counter is to be added to the address part of the instruction After this modification of the address part by the contents of one of the index registers or by the contents of the instruction location counter resulting in a modified address x1 position 6 of the instruction word is checked In the case the number in position 6 is 0 the instruction will be executed in the normal way with the modified address x 1 If the number in position 6 is 1 then the contents of location xl is read out of the memory and positions 6 to 12 of the contents of location x 1 replace positions 6 to 12 of the instruction in the instruction register address substitution Then the cycle starts again with a modification of the new address part by the contents of one of the index registeri or by the contents of the instruction location counter dependent on the number in position 12 of the instruction word and so forth The process ends when after an index register modification resulting in a modified address xn position 6 of the instruction word contains 0 Following this the instruction with address part xn is executed in accordance with the instruction list If the instruction to be executed is not indexable the modifications by the contents of one of the index registers are suppressed In the case the number in position 6 is 0 the instruction will be executed in the normal way If position 6 of the instruction word contains 1 only positions 6 to 11 of the contents of location x replace positions 6 to 11 of the instruction word in the instruction register Then again position 6 of the instruction word is checked and so on Instructions and apeed The 2002 contains 80 instructions 1 28 instructions for arithmetic opei ations for fixed-point and floating-point numbers shift operations and other specific operations In order to increase the calculating speed devices are provided which produce multiples of multiplicands and divisors Special attention -20 - has been paid to the built-in unnormalized floating-point ar thmetic so that roughly speaking the precision of the results is about the same as the minima3 precision of the two operands 2 12 jump and other control instructions one of whi z the so called subroutine jump UNT unterprogramm is executed as follows- supposed the instruction UNr 3 is stored in memory location a then a I is stored automatically by this instruction in memory location 3 precisely in positions 7 to 11 all other positions being reset to 0 the next instruction to be performed will be taken from memory location 3 1 The last instruction of a subroutine is a fine example of address substitution The unconditional jump instruction SPR springe for switching the control to memory location a 1 of the main program has the form SPR 103 where position % address substitution is equal to 1 3 9 instructions for index register operations including Jump instructions dependent on the contents of an index register 4 4 instructions for a transfer of data between drum and core memory in blocks of variable length 5 9 instructions for punched paper tape input and output 6 9 instructions for punched card input and output 7 7 additionai instructions for magnetic tape equipment 8 2 instructions for a cathode ray tube output unit The sign of an instruction is used for program testing Normally the sign of an instruction is not considered by the control unit After pressing a button on the control desk instructions with a positive sign are executed as usual instructions with a negative sign initiate the following procedure 1 The contents of the instruction location counter is stored in memory location 0 2 The next instruction in sequence will be taken from memory location 1 In memory location 1 for example there may be the start of a program printing the contents of the instruction location counter of the accumulator a s f tracing After the execution of this program the control can be switched back to the main program by a jump instruction SPR 1 0 using address substitution All operation cycles of the 2002 are integral multiples of the basic machine cycle which requires 90 microseconds One basic machine cycle is equal to the time interval necessary for adding the contents of a memory location to a number in the accumulator including the reading out of the addend Reading out of an instruction from the core memory its interpretation by the control unit and the address modification by index registers are performed in one basic machine cycle Each address substitution requires another 90 microseconds The execution of an instruction is not normally controlled by the central control unit but by separate circuits of the arithmetic input and output unit a s f That means that normally reading interpreting and address modifications of the next instruction are performed by the central control unit simultaneously to the carrying out of the preceding Instruction provided the execution of the preceding instruction requires more than one basic machine cycle For this reason the numbers bnted under Operational Speed do not include the 90 microseconds for reading interpreting and address modification by index registers of the Instruction The next instructions are even carried out during the execution of a preceding one If this is possible i e if the control circuits for carrying out the next instruction are not used by preceding instructions Under certain assumptions concerning the distribution of instructions of a program 25% additions 25% multiplications and 50% Instructions for organization a s o the average speed of the 2002 amounts to 2200 operations per second for fixed point and 1850 for floating point calculations - L 21 - iipu'rat ionsI SIproeda iii'lOttiIII Adde Multipl re'atingI OAR ft sod point III t il ' 01101 41d Ot 11114 rop potndlm II h inMaedoby The lastr Istrution ni The averae't i acestime'w f ordto retly reduer edbowhen transfrring blpoks of iooreasinable length v the'io adruo memoryki is% The' drumgt fey a atrqinct tiitsi 23 milliseconds Othe drum clock frequency St0O ke The avivrag' access tMe' ito thlt firat wiird of a block a about 19 milliseconds With atword transfer tit 90 iirs'ecdth tranofetr time of a block of n words Is 19 ý n -0 09 millisocondsm The average trunsfer thime At 1t 00 words from the drum memnory to the core' memory Its thereifore' 109 millise'conds In pt-Output Input and output data are' handled by 1%1a8ns of punchod patter tape' puint hiod card ia-i ii peitic tape equipment In the' case' of punche'd paper tapeo data cvan beMroiad itob the machine one character per Instruction with a speed up to 200 characte'rs per se'cond Results are puniched onl patpor tape at a speed of 50 characters per arcond A special buitlt-In input ro utine allows thf iaput of data without using a stored program The punched card control unit contains a core' buffer storage for e'ach punched card Input reap output The buffer storage' has a capacity of 80 x 13 96n bits thlt valumnne' rows of the' buffer being the equivalent of the' columnis row of the punched card The' nei'l cutrit is read into the input huffer start gre the ecntonts of the output butffer storage is punehed into the' next card by separate control circuits while calculating A special column numbered 0 ist provided In the input buffer storage' Any of the' 980 points of a vard can be' wired Into any p ositiont of column 0 Data Is read out of the' input buffer columnuwise beginning with columni 0 continuing with column 1 2 3 Up to 12 consecutile' colunmns can be re'ad by' one' Inttruction the' giumi' c of 'iolumns as well as the buffer unit are' indicated by thle addresp part oif tilt read insrtiuction The execution time for reading 12 columns is 540 microsecenids Data tit read into tht Accumulator There are two groups of instructions for reading numerical and aatphanume'rivai characters alphanumerical characters being represented by two decimtal digits The' output buffer In filled columninise In a similar way If a printer it ito be' connectod with anl output buffer the number of columns can be Increased The magnetic tape equipment is attachedt to tho 2002 in a nimilar way The' tape' control unit contains one or two buffers with a capacity of 1280 x 6 bits ona' charavter per colunmn The net record on tape is read Into tho buffer storage' ti e contents of thle' buffer s'torage' is written on the tape by read write instiuctions by separate c ntrol circuits wvhile calculating If two buffers are installed two records can be transferred simultaneously to or front twol tape' units Up to 10 tap e units can be connected with the tape' control unit The Instructions for readi ig datqL out of the tape buffer Into the accumulator and for filling the tape buffer ft am the accumulator a'e the same as for the punched card buffers Thle execution time for readling 12 columns Is 270 micro econds Because of the' length of tile tape buffer four block traicafer Instructions are provided which transfer thle next n columns of tape buffer k into the car ý memory beginaiing with memory location x resp In like mianner the' datA stored in the core memory beginning with memory location x transfers into the next n columns of tape buffer k Two instructions serve the transfer of alphanumerical and two instructions the transfer of numerical data Special instructions permit the sensing of the conditions of each input and output unit Thus for example each input and output buffer is equipped with a flip-flop which can be sensed and is 1 during the reading In of the next record punched card or during writing the contents of the buffer on tape or punched card This makes it possible to continue the program up to the very moment the required input or output buffer gets free The simultaneous carrying out of several input and output onerations is possibIP because of independent control circuits for each input and output buffer Certain applications necessitate a digital-analogue conversion of results A cathode ray tube unit permits the analogue display of output data 3 decimal digits representing the x coordinate and 3 decimal digits the y coordinate Generation of axes and diagonals is possible by one instruction The cathode ray tube unit can be furnished with a camera unit the transportation of the film is effected by the central processing unit SOCIETE D'ELECTRONIQUE ET D'AUTOMATISME - COURBEVOIE FRANCE The SOCIETE D'ELECTRONIQUE ET D'AUTOMATISME has just completed the installation of 2 large electronic sets for data processing statistics and management The first is at the INSTITUT NATIONAL de la STATISTIQUE et des ETUDES ECONOMIQUES in the establishment of general type statistics and various theoretical work The second is fcr accountant work statistics and operational research at the COMPTOIR FRANCATS des PRODUITS SIDERURGIQUES Their order was placed in October 1956 and the machine was delivered October 24 1958 and was put at the client's disposal on November 20 These two installations consist mainly of a general purpose digital computer CAB 3030 with a recorded internal program and three or four magnetic tape equipments CAB 3030 The 3030 works with a fixed point or floating point by program The machir word is 32 binary digits including one for sign and one for parity Negative numbers are represented by their 2's complement in order to obtain immediate notification of overflow Instructions are single address-except for those involving accumulators which are addressable by a second address The machine is serial with 4 accumulators and a parallel binary multiplier of one order and the finding of the next are carried out si iultaneously The execution Storage coi sists of magnetic drum with 128 tracks of 128 words each 16 384 words a quick access ferrite core store with a capacity of 1 024 words Transfers between the dr'im and the core store are performed automatically A parity digit is used for control in the transfer of numbers and instructions between ýhe different elements Inputs are Ferranti photo-electric punch tape reader 200 or 400 characters per second Fiexowriter electric typewriter and magnetic tape equipment 30 groups of 32 words or 160 characters per second Outnuts are Flexowriter electric typewriter S E A high speed punch at 45 characters per second S E A electronic printer Numerograph on microfilm 2 000 characters per second andi anagnetic tape equipment The 30310 has 3600 valves and 35 000 diodes Power required is 30 kilowatts Numerograph The first Numerograph has recently been set up at the Societe MONSAVON L'OREAL it is a fast electronic printer which enables the immediate printing on standard mirro ilm -c the output from an electronic computer at a rate which can reach 2 000 characters per second 93- A 6-bit code at the input controls two types of ordiers Writing and Auxiliary Auxiliary orders Capitals - Lower case - Advance - Return - Tabulation - Slacing Colour change The Capital and Lower Case orders enable the selection to be made from the group corresponding to a given cathode ray tube where the characters originate The character writing orders as well as Advance Tabulation and Spacing determine the position on the line A special electronic counter performs the same part as the carriage on a typewriter The position of each character is set on a special cathode ray tube based on the information received from the counter and a camera placed in front of this tube registers the results The Spacing order moves the film forward The Change colour order changes standard type to italics through a simple electronic anamorphosis Writing orders The coded information positions the Character cathode ray tube spot and in this way selects the corresponding character drawn on a 'ansparcnt sighting screen At the same time the photographed tube spot is pl-ree- on thc l nc - hich Is written at the place indicated by an Internal counting and tabulation system in the Numerograph The 2 spots synchronously sweep the rectangle surrounding the character to be reproduced A photoelectric cell placed in front of the character cathode ray tube transmits to the photographed cathode ray tube In front of which a camera photographs the characters as they are inscribed At present groups of 42 Capital and Lower Case characters are available This can be increased to 50 for each group The number of groups could subsequently be increased to 10 each approximately 40 characters At the present both straight and Italic characters are available The introduction of auxiliary orders Index and Exponent could subsequently be envisaged with changes in size involving whole groups of characters which would give with the previous combinations 3 sizes of characters Printing rate is 840 microseconds per character Line spacing requires 10 milliseconds These times could be subsequently reduced to 560 microseconds and 5 milliseconds with a forward film speed limit of approximately 150 lines per second Most errors are automatically detected with the exception of character alignment faults framing of characters on the line which can be corrected by a simple and rapid adjustment TAPE MERGING EQUIPMENT - ULTRA ELECTRIC LIMITED LONDON ENGLAND Ultra Electric's first entry into the Data Processing field is the fully-transistorized Tape Merging Equipment which was shown at the London Computer Exhibition This equipment is primarily intended to combine the information recorded on two punched paper tapes automatically producing a new tape which contains the result of the merging operation In a computer installation fitted with paper tape equipmeat the Tape Merging Equ ipmen makes it possible to exploit some of the powerful data-processing techniques which have been evolved for magnetic tapes so that the time of using a computer can be much reduced and It becomes practical to do small and medium data-processing tasks Further use of this equipment in conjunction with small computers opens up new uses for them in data-processing and it has applications in the building-up of programmes A typical application is in up-dating an inventorv file The two Input tapes are the brought-forward file and a tape containing amended Items the latter having been prepared on the computer The amended items tape also contains programme instructions which cause the equipment to copy items from -PIher of the two input tapes or to ignore an unwanted item on the 1rought-forward file tape In 'his way an up-dad c-at-rrorv ard file is p'reared as an off-llne i-p ratin The equipment can also tunction as a high-speed tape copier or carry out these two functions simultuineouely w rAorwparator or can ELECTRONIC COMPUTER SZ ON - THE WEIZMANN INSTITUTE OF SCINNCE REHOVOTH ISRAEL WEIZAC' The memory capacity of WEIZAC see Digital Computer Newslotter January t14t5el-au % on boosted by the installation of tuu magnetic tape handlers using Potter 905 tape transports The desagn ol Lav tape circuits is conservative character repetition rate being 5000 per second Theme tape units were used heavily during 1958 doing mostly high-order matrix work Present plans call for the installation of two additional tape units during 1959 doubling the character rate An off-line device has been built for the print-out og data from magnetic tape onto paper tape via a magnetic drum This unit is now undergoing final tests Total Code sChecking Month Production MonthDevelopment Hr Mina Hr Mins Scheduled Computation Time Hr- Idle Time Mina H -- Engineering and Mine Hr Unscheduled Breakdowns Mina Hre Mina Oct 57 78 48 252 02 330 50 5 23 84 40 20 05 Nov 87 43 309 09 376 52 5 48 70 46 8 03 Dec 108 37 332 27 439 04 5 46 58 49 14 14 Jan 58 52 48 408 47 461 33 5 45 79 14 30 52 Feb 53 22 378 33 431 55 - 58 76 26 44 28 March 54 57 491 04 548 01 2 42 77 55 44 42 April 51 19 568 48 618 05 1 51 53 32 31 47 May 88 49 574 21 641 10 1 49 65 16 20 25 June 57 18 532 12 589 30 - 05 52 50 52 24 July 42 25 554 31 598156 3 28 55 27 48 42 Aug 41 45 561 38 603 21 - 23 60 01 57 30 Sep 86 57 416 23 483 20 2 25 40 07 32 58 Oct 45 51 617 30 863 21 2 28 40 45 24 18 Nov 80 47 547 41 608 28 4 56 62 20 42 58 Dec 88 47 527 13 616 - 1 5 66127 48 48 936 1 7070 15 8008 26 520 1 45 137 924 35 Recent problems They have solved for the first eigenvalue of a determinant of order 1078 The problem arose inconnection with the determination of the energy level of the ground state of Helium They have also been engaged inthie development of methods of carrying out -25 - algebraic operations on WEIZAC The need for this approach arose kn an investigation on the propagation of a seismic pulse in a layered elastic half-space A recent trend in their work has been to operate with invisible equations the equations be•ome so long that it Is impossible to write them down so that WEIZAC increasingly operates with equations which it generates itself COMPONENTS HIGH SPEED PRINTER - A B DICK CO - CHICAOO ILLINOIS Time Inc has awarded a contract to A B Dick Company of Chicago IPlinois for 2 address-label printing systems each capable of printing in excess of 130 000 1 x 2-3 4 labels per hour per system from digitally coded magnetic tape Each system will include an A B Dick Model 910-1 label printer and a Model 940A tape reader and buffer unit The printer unit employs A B Dick Company's recently announced VIDEOGRAPH highspeed printing process see Digital Computer Newsletter July 1958 Principle element of the process is a special CRT having a dense matrix of fine wires permanently bonded through the face-plate instead of the conventional phosphor screen Deflecting and modulating the electron beam across the inner side of the matrix while simultaneously transporting dielectrically coated paper past the face of the tube deposits latent charges on the paper which are then developed by dusting with a powder The visible images are then fixed on the surface In employing this process with digital input systems a specially developed character generator is utilized which converts pulses into the alphanumeric video waveforms required to drive the electrostatic printing tube at writing rates that may be in excess of 20 000 characters per second In the Time Inc equipment the tape reader and buffer unit TRBU will fully edit the information provided on the magnetic tape input and provide a continuous flow of data to the printer The unit will employ a 2184 character core buffer to receive the edited information from which the city and postal zone codes are read into a 144 character magnetic storage unit for recirculation into the output for the successive series of labels directed to the same city and zone Delivery of the two printing systems to the magazine's Subscription Service Division in Chicago is scheduled for June 1960 Major subcontractors include Stanford Research Institute Cook Electric Company and Telemeter Magnetics Inc HIGH SPEED DIGITAL PLOTTER - LOCKHEED MISSILES AND SPACE DIV SUNNYVALE CALIFORNIA The High Speed Digital Plotter developed by the Computer Research Department of the Lockheed Missiles and Space Division accepts magnetic tape from IBM or Remington Rand Univac Computers and records the output on electrolytic facsimile paper The maximum plotting rate is in excess of 4 000 data points per second in addition to coordinate lines which are automatically generated Data points are recorded on facsimile paper by the passage of current from selected styli which are arranged in a line perpendicular to the direction of paper travel Spacing between styli is 0 01 inch The paper feed rate and electronic timing allow marking of adjacent points 0 01 inch apart perpendicular to the stylus array Thus accuracy is 0 01 inch for both X and Y coordinates Many curves including a complete coordinate system can be written simultaneously with this plotter The use of the high speed plotter has already proven of great value in speeding data reduction It is now possible to produce completely annotated digital plots within a few hours following - 26 - I a test flight With the great speed and flexibility of the plotter it should prove to be an increasingly valuable tool for reducing data SYSTEMS COMPONENTS - PACKARD BELL COMPUTER CORP LOS ANGELES CALIFORNIA - The Impact Prediction Systems for Vandenberg Air Force Base has been delivered This system is based on the conversion of shaft position information into digital form computing certain functions and converting the output to voltages for display on analog plotters A paper describing this system is available for anyone who wishes to write for it Certain systems are being made in which a Bendix G-15 general purpose computer is being tied to an analog computer In one typical system there are the channels of analog-to-digital information being multiplexed into the G-15 and three channels of digital-to-analon information being transferred back to the analog computer The cost of the coupling unit is approximately $25 000 This same technique may be used on other computers and on more elaborate systems SERIES RB STORAGE UNITS - TELEMETER MAGNETICS INC LOS ANGELES CALIFORNIA - The Series RB random access data storage units have been announced by Telemeter Magnetics Inc These memory-buffer units offer addressable random access sequential access or a combination of both as desired and have a combination of features not previously realized High Speed 200 kc operating rate Long Term Reliability Solid state active elements conservative derating of components advanced design and manufactqring techniques Wide Range of Capacities From 256 to 1024 words of from 4 to 20 bits per word Random or Sequential Access Addressable storage registers for regenerative read write or buffer operation in a sequential non-regenerative mode Binary or Decimal Either binary or binary coded decimal addressing Variable Input Output Will accept pulses or levels of either polarity and input may be changed during operation Output Is equally versatile Accepts and emits single-ended or double-ended information Speed Load or unload a word all bits in parallel in 5 microseconds Random access with regenerative storage - complete cycle in 10 microseconds Operating Mode Sequential loading and unloading Random access addressing for loading and unloading or regenerative read write Operations can be intermixed in any manner desired without loss of speed Control Levels and Signals Input and output levels may be single ended or double ended and a ONE may be represented by -5 or 5 volts as desired Input pulses must rise to between 2 and 10 volts from a quiescent level of -5 volts Rise time between 0 2 and 1 0 microsecond Output pulses rise to 5 volts from a quiescent level of -5 volts At the top of the pulse 25 milliamperes are available Reference levels of 5 volts are available at the output of the unit 27- Clearing Electronic clearing is provided and address register may be cleared to the all ONE or all ZERO state Power Required Nominal 115 volts 60 cps less than 250 watts Satisfactory operation is obtained from supply voltages between 100 and 130 volts Dimensions Size depends on capacity Largest Series RB unit approximately 30 inches high by 14 inches deep All units are supplied for relay rack mounting Environment Reliable operation between 00 and 55 C Unaffected by humidity Low Cost Series RB random access storage units are priced below most memory units designed to do only a portion of the job these memories will handle MISCELLANEOUS CONTRIBUTIONS FOR DIGITAL COMPUTER NEWSLETTER The Office of Naval Research welcomes contributions to the NEWSLETTER Your contributions will assist in improving the contents of this newsletter and in making it an even better medium of exchange of information between government laboratories academic institutions and industry It is hoped that the readers will participate to an even greater extent than in the past in transmitting technical material and surgestions to this Office for future issues Because of limited time and personnel it is often impossible for the editor to acknowledge individually all material which has been sent to this Office for publication The NEWSLETTER is published four times a year on the first of January April July and October and material should be in the hands of the editor at least one month before the publication date in order to be included in that issue The NEWSLETTER is circulated to all interested military and government agencies and the contractors of the Federal Government In addition it is being reprinted in the Communications of the Association for Computing Machinery Correspondence and contributions should be addressed to GORDON D GOLDSTEIN Editor Digital Computer Newsletter Information Systems Branch Office of Naval Research Washington 25 D C - 28 - Government regulations distribution for your - TO for convenience to indicate Do not return in Please send toCode correct D the maintenance all periodicals rections Washington 25 I require lists this The necessary sheet 437 if your form below changes or presenx Office of NavaJ I j REVISE AS INDICATED BELOW I I listing Research Office of Navall I I I DELETE FROM DISTRIBUTION LIST I Please print or type From PRESRNT ADDRESS present address as it envelopes MUST be listed I is cor- C DIGITAL COMPUTER NEWSLETTER Code 437 Research Washington 25 D C I of up-to-date The complete appears here To CORRECT ADDRESS on the I I I II S INAU DATE I