Sign in

Stored-program computer design

1945Turning point

John von Neumann describes storing programs in memory alongside data, the architecture of nearly all computers.

The story

On June 30, 1945, the mathematician John von Neumann's 'First Draft of a Report on the EDVAC' began circulating among scientists. It described the logical design for a successor to ENIAC with one crucial difference: the computer's instructions would be stored as numbers in the same memory that held its data. Instead of rewiring a machine to change what it did, you would simply load a new program. The report laid out the parts of such a machine: an arithmetic unit to do calculations, a control unit to fetch and carry out instructions one by one, a memory, and devices for input and output. This blueprint is known as the von Neumann architecture.

The idea grew from the ENIAC project. ENIAC's builders, J. Presper Eckert and John Mauchly, had already realized that storing programs in memory would remove the days of plugging cables, and they had discussed designs with von Neumann, who joined as a consultant in 1944 while also working on the atomic bomb. Alan Turing's 1936 universal machine, which read its instructions from the same tape as its data, had established the theory. Von Neumann's contribution was a clear, general description, written in an abstract notation partly borrowed from McCulloch and Pitts' model of neurons.

The report caused a lasting dispute. Herman Goldstine circulated it with only von Neumann's name on it, and Eckert and Mauchly felt their ideas had been taken. Because it was widely distributed, it also counted as a publication, which undermined patents on the stored-program concept and helped put the idea into the public domain.

Machines based on the design soon appeared. The Manchester Baby at the University of Manchester ran the first stored program on June 21, 1948. EDSAC at Cambridge, built by Maurice Wilkes, began regular service in 1949, and EDVAC itself followed. Storing programs as data also meant programs could create or modify other programs, which made compilers and programming languages possible. Nearly every computer since, from mainframes to smartphones, follows this basic plan, though the single channel between processor and memory, the 'von Neumann bottleneck', remains a limit engineers work around.

Why it mattered

  • Programming languages such as FORTRAN became possible because programs could be stored and translated by other programs.
  • Fast, reprogrammable computers enabled numerical experiments such as the weather simulations that revealed chaos theory.
  • Commercial computers from UNIVAC and IBM adopted the design, starting the computer industry.
  • Software emerged as a separate product from hardware, eventually becoming an industry of its own.

Drafted by AI (Claude) from the sources above and checked against them by AI; no historian has reviewed this page. Spot a mistake? Suggest a correction. How pages are made

Community notes

Loading notes…

Learn more ranked by votes