![[the-most-amazing-coincidence-aiken.jpg]] ![[family-tree-of-machines.png]] ## 1939-1944: Harvard Mark I The Lawnmower. 1939: The Harvard Mark I. Howard Aiken. Completed in 1944. The Mark I was not electrical. 51 by 8 foot panel. Controlled by mechanical switches. Driven by a four horsepower motor. For reference, a standard walk-behind lawnmower has about a 2-7 horsepower motor. So this thing is a lawnmower with a 400 square foot interface made of switches and dials. Not really a computer in the Turing complete sense. Closer to a player piano. Could do 3 additions per second. Did base 10 addition to 23 significant figures. ## 1944-1949: EDVAC The von Neumann architecture. 1944-1949: EDVAC. Electronic Discrete Variable Computer. US Army offers a contract to University of Pennsylvania to make this. Various sources say the contract was signed in 1944, 1945, or 1946. Was delivered in 1949. Original team was Herman Goldstine, John Mauchly, and John νon Neumann. Used an ultrasonic mercury delay line memory, which could store 1,024 44-bit words. Binary rather than decimal, and designed to be a stored program computer because von Neumann was working on it and obviously. EDVAC's average addition time was 864 microseconds and its average multiplication time was 2,900 microseconds. ## 1945: ENIAC > _The ENIAC computer. Well, we can call it a computer. It's a calculator. It's not really universal._ > -Andrew Appel, Turing, Gödel and Church at Princeton in the 1930s The First Egypt. 1945: ENIAC. Electronic Numerical Integrator And Calculator. J.W. Mauchly and J.P. Eckert of the University of Pennsylvania. 18,000 tubes and 1,500 relays. Commissioned by the Ballistics Research Laboratory at the Aberdeen Maryland proving ground. Was the electrical analogue of the Mark I, but ran several hundred times faster since it was not a lawnmower. The first instance of "Best developers quit because company says everything you do is ours" was in 1946, right after the literal damn ENIAC. > _In early 1946, months after the completion of ENIAC, the University of Pennsylvania adopted a new patent policy, which would have required Eckert and Mauchly to assign all their patents to the university if they stayed beyond spring of that year. Unable to reach an agreement with the university, the duo left the Moore School of Electrical Engineering in March 1946, along with many of the senior engineering staff. Simultaneously, the duo founded the Electronic Control Company (later renamed the Eckert-Mauchly Computer Corporation)._ > > -The Dynamic Read Writable Free Encyclopedic Repository of the Modern State of Human Knowledge. ## 1947-1951: UNIVAC The First Exodus. 1947: UNIVAC. UNIVersal Automatic Computer. First "Developers quit and start company which doesn't quite work but sort of" in computing history. Eckert and Mauchly of ENIAC fame quit University of Pennsylvania, form their own computer company, and start building UNIVAC, which has a way cooler name than any of the previous machines. Naturally they ran out of funds, but then got $500k in funding from a gambling machine company, possibly the first intersection of computing and gaming. Gambling company takes over and then funds run out again. Sells the Eckert-Mauchly company to James Rand of Remington Rand, a company that made typewriters and guns. UNIVAC finally shipped in 1951, over a year late, with six more on back order. ## 1946-1949: EDSAC The First Stdio Library. 1946-1949: EDSAC. In 1946, Maurice Wilkes returns from Pennsylvania to Cambridge and begins work on the Electronic Delay Storage Automatic Calculator. Inspired by ENIAC but 1/5 of the size. Performed its first fully automatic calculation in 1949. First electronic machine with high speed memory and I/O devices. Within a few years, the EDSAC's standard library had over 150 subroutines. ## 1947: The Manchester Baby 1947-1949: The Manchester Machines. - Sources conflict on details of the MADM, Manchester Baby, and the Manchester Mark I, and whether Turing was involved. - Around the same time as EDSAC, in Manchester, a team consisting of M.H.A. Newman, F.C. Williams, I.J. Goode, and ALAN FREAKING TURING get together to build the Manchester Automatic Digital Machine prototype. - I.J. Goode had worked on the Colossus code breaking machine during the war. And Alan Turing is obviously Alan Turing. - The MADM prototype was built in 1948 and the full machine ran its first program in June 1949. Introduced the index register and pagination. - Max Newman arranged for the "acquisition of war-surplus supplies for its construction, including GPO metal racks and '...the material of two complete Colossi' from Bletchley." - First computer to execute a program stored in addressable read-write electronic memory. ## 1948: SSEC The First "80 Column Line" 1948: SSEC. Put into operation at IBM headquarters in Manhattan in 1948. Cleverly placed at street level behind large plate glass windows so passers-by could see it working. John Backus got his start working on the SSEC in 1950. ## 1951: Whirlwind The First Real Time System 1951: MIT Whirlwind. First machine with magnetic core memory. Was used by a young Doug McIlroy in 1954. One of the "bright boys" on the project was Kenneth H. Olsen, who eventually founded Digital Equipment Company (later Corporation.) In a _student paper,_ Kenneth Olsen suggested using transistors instead of vacuum tubes. > _Until this point, all computers constructed were dedicated to single tasks, and run in batch mode. A series of inputs were set up in advance and fed into the computer, which would work out the answers and print them. This was not appropriate for the Whirlwind system, which needed to operate continually on an ever-changing series of inputs._ > > -The Dynamic Read Writable Free Encyclopedic Repository of the Modern State of Human Knowledge. ## 1952-1954: The 701 and 704 The First "High Level" Languages. Coming from the SSEC in 1950, around this time John Backus starts making noise. 1952: IBM 701. John Backus develops "Speedcoding" to help him write code for the 701. 1954: IBM 704. Backus assembles a team to define and implement FORTRAN, the first high level language that was broadly used. Fortran code can still be found in the Scipy project in 2025, literally _70 years later._ Later in 1958, John McCarthy would begin creating LISP at MIT, with the goal of getting an AI programming language that could run on the IBM 704.