Richard C. Lesser's Recollections
The Cornell Computing Center - the early years, 1953 to 1964
Computing at Cornell had its beginning in the spring and early summer of 1953. Professors Robert J. Walker and J. Barkley Rosser of the Mathematics Department and a small group of faculty formed an advisory committee to discuss the establishment of a computing center that would that would offer services to academic and research staff. The decision to go ahead was made by Dr. T.P. Wright , the Vice President for Research and Provost Hill. The Cornell Computing Center was to be set up as a part of the Dept. of Mathematics with an advisory committee representing interested groups on campus. At this time I was on the staff of the Center for Statistical Services and Scientific Computation at MIT, which provided such services to Institute faculty. The "Scientific Computation" area was my responsibility. There was no MIT Computation Center as yet on the horizon. That summer I became acquainted with a Cornell mathematician, Professor Mark Kac, who was a visiting professor for the summer at MIT, and he recommended me to Professor Walker as a candidate for director when the Cornell center was formed. I visited Cornell in August 1953 and was shortly thereafter offered the position of Director of the Cornell Computing Center.
The CPC years 1953 -1956
The new center had two missions - first, to make computing power available to the faculty and graduate students, and second, to assist the campus in gaining computer knowledge and literacy. It is hard to believe today that there was no use of computers in the classroom. Before I moved to Ithaca in Sept., 1953, we placed an order for an IBM Card Programmed Electronic Calculator (CPC) for delivery in the fall. It should be noted that the center was to support itself by selling time to university departments, and, to assuage administration fears of a deficit, the Cornell Aeronautical Laboratory in Buffalo guaranteed a portion of the budget. The estimated budget for the period Dec. 1, 1953 to June 30, 1954 was a staggering $19,275.
|
|
Lesser and others at CPC in Rand Hall (click on picture to see a larger image) |
The time before the delivery of the CPC was spent meeting with faculty and giving talks to any group that was interested. In preparation for the equipment, I spent untold hours wiring and testing the board that controlled everything. The CPC was a composite of three separate machines that were cabled to function as a single unit. Card input and printed output was through an IBM 418 accounting machine (tabulator, or tab for short), card output was via an IBM 527 reproducing punch, while calculations were performed by an IBM 605 electronic calculator (which was a successor to the widely used IBM 602A calculating punch). Decks of cards with instructions were fed into the tab, (at the rate of 100 to 150 cards per minute), calculations were performed, intermediate results punched out onto cards, and reentered in new instruction decks for further work. One instruction card could perform an operation such as [AxB+C=D]. Sometimes the intermediate results were collated (appropriately, on an IBM 077 Collator) and iterated until the desired result was achieved. To invert a 20x20 matrix, for example, took 20 iterations and several hours, longer if a deck got dropped or out of order in the process. Despite its clumsiness, with ingenuity and perseverance we could do quite complicated computations with the CPC. For example, a square root was calculated by iterating a formula until acceptable convergence was achieved. The internal data storage was limited to registers in the tab and the 605, a problem that was somewhat eased by the arrival of a new storage unit, the IBM 941A. As I sit here with a gigabyte of hard disk, I remember how exciting it was to use the 941 with its capacity to store 16 ten digit numbers (with sign!). To those who think extension conflicts are tough to resolve, I finally tracked down why an otherwise reliable computation dropped a "1" about every three months. (It turned out that a pilot selector (relay) in the tab dropped out for 10 degrees of its 360 degree cycle and any "1" passing through at that time was lost).
The Center was housed in the east end of the third floor of Rand Hall in spacious, if ancient, quarters. Our nearest neighbors were Buckminster Fuller and his students building geodesic domes. The initial staff was myself, Ann Waymeyer as secretary and keypunch operator, and Dorothy Hartman, as programmer, key punch and computer operator, and in the first summer of 1954, a high school student, Margaret Hunter as jack-of-all-trades. In addition to the CPC, we had keypunches, a verifier, a sorter and the collator.
|
|
Lesser, McNair, Chu watch the CPC make maps from photos
(click on picture to see a larger image) |
Finally, on December 15, 1953, at 8:00 AM, the CPC was powered up for the first time. (The time sheet for the first week of operation has been preserved.) Much of the following weeks were spent in testing the boards and the first programs. It should be noted that the main board (the tab board) controlled most of what went on in the CPC. It was about 3 foot square and was piled high with over 3 inches of wires, color coded as to length. Making changes or additions was a daunting procedure. Making a board diagram was a must, in order to be able to follow what was going on. I have had no liking for documentation of programs to this day as a result.
Our first customer was Professor Lyman G. Parratt of the Physics department. In the first year many of the jobs were for faculty in the state colleges of Agriculture and Home Economics, mostly statistical in nature, such as correlations and analyses of variance. Early customers were Professors Urie Bronfenbrenner of Child Development and Family Behavior, Walter Federer of Vegetable Crops, M. T. Vittum of the Geneva Agricultural Experiment Station, Richard Bersohn of Chemistry and Arthur McNair of Civil Engineering. We charged customers at the rate of $25 per hour.
The 650 years 1956 -1959
During 1954 and 1955, demands for the Center's services increased steadily. The staff continued to grow with additional programmers and support staff. In late 1954, we placed an order for an IBM 650 magnetic drum computer for delivery in early 1956. The acquisition of the new machine was furthered by a $50,000 grant from the National Science Foundation, and a 60% rental contribution from IBM. This new machine would require more space and air conditioning ( but the staff would not) so a new home was called for. With the help of Dean S.C. Hollister of the College of Engineering and Prof. William Erickson, Director of the Dept. of Electrical Engineering, space was found on the first floor of the new EE building, Phillips Hall. In addition to the computer room, we had an auxiliary machine room and an office for the Director. Our first research (graduate) assistant, Virginia A. (Ann) Walbran, joined the staff.
|
|
Lesser at IBM 650 in Phillips Hall
(click on picture to see a larger image) |
The 650's drum (storage unit) was capable of holding 2000 words (numbers) of 10 decimal digits and sign. It was our first experience with a "stored program" machine and the days of running program decks over and over on the CPC were soon forgotten. Information was read into and punched out of the 650 via the IBM 533 Read-Punch unit and the output was printed on an IBM tab. Instructions and data were interpersed on tracks on the drum, and execution speed was dependent on placing data so that it would be available as the drum rotated. Initially, programming was in machine language and the process of making sure the data was available was called hand- optimization. This was a time consuming programming task, requiring knowledge of the rules for how soon the drum would rotate to a given location. As a program got larger, one had to keep track of available locations and be careful not to slow the run- time by requiring a full rotation of the drum before data could be accessed.
Fortunately, help was on the way in the form of new ideas, the interpretive and the assembly programs. The first of these was described in Technical Newsletter #11, published by IBM in March 1956. (It should be noted that Technical Newsletter #1 was the first {non-operating manual} literature published by IBM). This interpreter was titled "A Complete Floating-Decimal Interpretive System for the IBM 650 Magnetic Drum Calculator" and was developed by Dr. V. M. Wolontis at the Bell Telephone Laboratories. The 650 being a fixed-decimal machine, it was transformed into a "three- address, floating-decimal, general purpose computer, primarily suited for scientific and engineering calculations". The assembler was the Symbolic Optimal Assembly Program, the famous S.O.A.P., described in IBM's 650 programming Bulletin 1, dated May, 1956. In 650 programming, the operation (OP) code 65 stood for Reset and Add to the Lower Accumulator, abbreviated RAL for the purposes of the assembler. (RAL are also my oldest son's initials, giving me a lingering link to the 650). SOAP gave us the ability to use mnemonics for operation codes and for operands as well. Programming was performed according to SOAP's rules, and a program with the resulting OP codes and optimized addresses was assembled at the rate of 50 to 75 cards per minute. This program, when punched out, could be listed on the tab for debugging, or run back through the 650 for execution. Programming had advanced from card programming through machine language to a new level of ease. Our rate for sponsored computing was now $75 per hour.
The 220 years 1959 -1962
It seemed almost axiomatic, in the early days of computing at Cornell, that the capacity of the current equipment was exceeded every three years. So it was that in 1958 we began looking for a successor to the 650. Once again, a search for a new home for the Computing Center was necessary as we had overflowed the existing space available in Phillips Hall. After a long evaluation of available equipment, it was decided to order an Electrodata Datatron 220, (renamed the Burroughs 220
|
|
Burroughs 220 in Rand Hall
(click on picture to see a larger image; click here to see another view) |
The $100,000 renovation of Rand Hall for the 220 was extensive, as a large air-conditioning system was required for the last of the great tube machines ( magnetic core was just on the horizon). In addition, a bright yellow raised flooring was installed so that the multitude of cables connecting the components would be hidden. Installed in 1959, the 220 was a decimal computer with 5000 ten-digit words of storage. New features to Cornell were magnetic tape and paper tape systems, as well as a supervisory printer linked to the console. I claim the "honor" of being the last individual allowed to debug programs at the console. Input was by cards on an IBM collator, and, very occasionally, paper tape. Output was printed on an IBM tabulator, or punched by an IBM reproducer or on paper tape. We charged $150 per hour for computer time used by sponsored research projects.
Cornell was one of the founders of the original Burroughs User Group, the Cooperating Users' Exchange (CUE), of which Robert Gordon of Stanford was the first president, and I was the second. This group had large representation among Universities and military research facilities and, through the urging of this group, the next breakthrough in software was achieved. Burroughs Corporation, in early 1961 published and distributed the Burroughs Algebraic Compiler, " a representation of ALGOL for the Burroughs 220 Data-Processing System". As had board wiring and machine language programming, assemblers now dwindled in importance. We had been using the assembler CAP (Cornell Assembly Program), written in 1959/60 for the Computing Center by an undergraduate student, David J. Waks. With the availability of the CAP assembler and the BAC compiler, programming the 220 was greatly facilitated.
The next programming advance at Cornell came in the early 1960's, when Professors Richard Conway and William Maxwell of the Department of Industrial Engineering, and Robert J. Walker of the Department of Mathematics wrote an interpretive compiler called CORC, the Cornell Compiler, similar to the widely used BASIC, which was being developed at Dartmouth. Whereas BASIC had a highly structured, formal syntax, CORC was programmed in English statements and was easy to learn and use. Initially written for the 220, and later for the Control Data1604, it increased student usage greatly. Programs submitted before 5 o'clock would be compiled or run during the night, and results made available to the student the next morning. By 1963, approximately 1,000 students were using the computers for coursework. In 1962, the College of Engineering incorporated computer programming as part of the third semester of calculus, and, in the academic year, 1963-1964, 14 credit courses related to computing were offered on campus.
The staff continued to grow. Management additions included Professor Seymour Parter, Mathematics, as Associate Director in charge of research, and John W. Rudan, as Assistant to the Director (later to become Assistant Director and Director). In 1960, the staff consisted of 18 including 4 programmers.
The 1604 years 1962 -1966
|
|
Control Data 1604-1604A
(click on picture to see a larger image) |
The three-year cycle continued, as Cornell ordered a Control Data 1604-1604A computer system for installation in the fall of 1962. The 1604 was 50,000 times the speed of the original CPC. This machine, valued at well over a million dollars, was again financed by a National Science Foundation grant as well as a substantial educational grant by Control Data. The 1604 was installed in a new computer room contiguous to the 220, also air-conditioned with raised flooring. Student assistants, both undergraduate and graduate, continued to provide much of what we now call systems programming. Some of these were Philip Kiviat, George Petznick, John Behrenberg, and David Bessel. With the acquisition of the 1604, the Center expanded into the second floor of Rand Hall to provide storage space and offices for additional staff. Assistant Professors Shayle Searle, Dept. of Plant Breeding, and Sidney Saltzman, Dept. of Industrial Engineering and Administration joined the staff as Research Associates. By the time I left in 1964, the Cornell Computing Center full and part-time staff numbered about 50. One of my last efforts at Cornell was to effect the sale of the Burroughs 220 to a former customer, John Middlebrook, a consultant to General Foods. I believe the sale price was $75,000, not bad for an obsolete, tube-type machine!
By 1964, many Cornell Departments were beginning to acquire computers to serve their own research and administrative needs, as were many of our former, outside Cornell, customers. Some of these were the University of Rochester Institute of Optics, University of Syracuse, GLF, Inc. (now Agway), Corning Glass Works, the GE Advanced Electronics Center, and optical companies in Rochester.
When I arrived at Cornell in 1953, Professor Kac had warned me it would be difficult to leave, so stimulating was the university and so beautiful was the environment. In 1964, that time had come, and I left in September for Albany and the State University of New York. I hope that these remembrances will be of interest to Cornellians, past and present, and that my memory has not admitted of too many errors or omissions. These early years were ones of trial, error, and discovery, challenge and fulfillment, and the problems we faced and solved (with limited tools) would be incomprehensible to those who practice "computing" today.
Richard C. Lesser,
Fripp Island, South Carolina
August, 1996
Bibliography.
Notes on the Electronic Calculator 604, Richard C. Lesser, 1953.
Notes on the CPC Model 2 and Component Machines, Richard C. Lesser, 1953.
Memorandum to Advisory Committee, R.J. Walker, October, 1953.
Memorandum to Anyone Interested in Machine Computation, R. J. Walker and R. C. Lesser, Cornell Computing Center, November, 1953.
Report to Associates, Number 4, Cornell University, January, 1954.
IBM Card-Programmed Electronic Calculator, Principles of Operation, IBM Corporation, October, 1954.
The Cornell Engineer, Vol. 20, No 3, Cornell University College of Engineering, December, 1954.
IBM Technical Newsletter No. 11, IBM Corporation, March, 1956.
IBM 650 Programming Bulletin 1, IBM Corporation, May, 1956.
Research at Cornell, The Annual Report of the Vice-President for Research 1955- 1956, Cornell University, May, 1956.
An Introduction to Coding the Burroughs 220, Burroughs Corporation, December, 1958.
Research at Cornell, The Twelfth Annual Report of the Vice-President for Research, 1959-1960, Cornell University, June, 1960.
CAP Manual, Cornell Computing Center, July 1960.
Cornell Alumni News, Volume 63, Number 7, December, 1960.
Burroughs Algebraic Compiler, Burroughs Corporation, January, 1961.
CORC, The Cornell Computing Language, Cornell University, September, 1963.
In excellence and diversity..., Cornell University, Undated.
Brochure, The Cornell Computing Center, Undated.
Richard C. Lesser, August 1996