The IBM pluggable sequence relay calculator

The first two machines of this type were built during the war for the Aberdeen Proving Ground, were delivered in December 1944 and were in operation during the last eight months of the war. In November 1945 they were returned to the factory for the provision of increased storage capacity and other improvements. At that time three additional machines of the improved type were built—one for the Naval Proving Ground at Dahlgren and two for the Watson Scientific Computing Laboratory at Columbia University. The latter are available to scientists engaged in basic research. The machines were designed and built in the Engineering Department of. the IBM by C. D. Lake, B. M. Durfee, and D. R. Piatt. The machine reads numbers from punched cards, performs a sequence of calculations on them by means of relay networks, and punches the results in cards. The calculating unit is the fastest relay unit that has been announced (November 1947); six-digit multiplications are performed in 0.15 second, which is about three times as fast as that of any other relay multiplying unit. The basic operations include addition, subtraction, multiplication, division, square root, and column shift. There are 36 storage and computing registers; mathematical tables with monotonie argument may be used. The reading and recording speed is greater than that of any other existing calculator. Cards are read and punched at the rate of 100 per minute, and it is possible to read four cards simultaneously and punch a fifth at this rate. In comparison with the IBM Sequence Controlled Calculator at Harvard, this machine is limited in internal storage capacity, number of significant figures, and flexibility of sequencing; on the other hand, multiplying speed is about 20 times as great, and the use of plugboard facilities and punched cards permits parallel operation (as distinguished from sequence operation), with further gain in efficiency. While the speed in performing the arithmetical operations is less than that of electronic machines, the over-all production on general computing is extremely high because of the many other factors involved. These factors include the use of automatic control panels for quick change from one problem to another; efficient card reading, punching, and comparing facilities; convenience of operation ; and the advantages of having twin machines in the same room. For a great many scientific problems, the internal storage capacity of all existing machines is quickly exceeded, and the over-all speed is determined by the facilities for reading and recording rather than by the speed of computing ; in such cases these calculators, which take full advantage of punched card storage, are unequalled in efficiency. A simple example of this type of problem is the manipulation of large harmonic series described in §6. The scope of the machine may be seen by comparing it with the standard IBM Calculating Punch (type 602). The principal differences include 149