EDVAC
The EDVAC (Electronic Discrete Variable Automatic Computer) was the first all-electronic computer (programmable, in the modern sense) designed in the US. (Depending on the exact definition of 'computer' used, it may have been preceded by the Automatic Sequence Controlled Calculator, or the ENIAC.)
It was conceived the University of Pennsylvania's Moore School in 1945, while work was still proceeding on the ENIAC; early consideration of how the ENIAC would be re-configured for a new problem indicated that that would be somewhat difficult. The notion of programmability was known already from the ASCC, and the Relay Interpolator, and that promised to speed up switching the machine to a different problem. However, the electro-mechanical devices used on those machines to read the program were an order of magnitude slower than the electronic calculation subsystem, as were similar devices used to hold intermediate results. Use of an electronic memory to hold both the program and intermediate results was the obvious answer.
John Mauchly and J. Presper Eckert had already begun to consider this change in direction, when Herman Goldstine, the head of the project, introduced the mathematician John von Neumann (who was familiar with the idea of programs, from his knowledge of Alan Turing's work on computability) to them. After further discussion, they rapidly came to rough agreement on the new approach, which almost every computer from that day forward has followed: a memory which merely stores things (unlike the accumulators of the ENIAC), and a separate subsystem which performs all the calculations. Von Neumann wrote up the results in his famous rough draft First Draft of a Report on the EDVAC - without noting who had contributed what, thereby igniting a controversy that was never really definitively settled.
The First Draft, dated June 30, 1945, was circulated fairy widely in that form by Goldstine, and was the inspiration for most of the first generation of computers; almost all computers now are descendants. The EDVAC itself was not very influential; Mauchly and Eckert left the Moore School shortly thereafter, over a dispute about patents, and their loss slowed down the project so much that the machine was not finally operational until 1952.
It used acoustic delay lines for its main memory (1K words in total, in 128 delay lines). It was constructed using vacuum tubes (although many fewer than the ENIAC; roughly 1/4th as many), and was a serial computer internally, using binary, unlike the decimal ENIAC. Instructions were 44 bits long, and contained a 4-bit opcode and four 10-bit addresses (two sources, a destination, and a 'next instruction'). It had floating point capability; 33 bits for the mantissa, and a 10-bits exponent.
Further reading
- Herman H. Goldstine, The Computer from Pascal to von Neumann, Princeton University, Princeton, 1972 - Goldstine was present during the initial design discussions
- William Aspray, John von Neumann and the Origins of Modern Computing, MIT Press, Cambridge, 1990
External links
- John von Neumann, First Draft of a Report on the EDVAC
- John Mauchly, J. Presper Eckert, A Progress Report on the EDVAC
