In the summer of 1944, long before ENIAC was operational, the eminent Hungarian/American mathematician John von Neumann, was introduced to the work being done on the ENIAC by Goldstine when they met by chance on a railway platform. Shortly afterwards, Von Neumann became a consultant on the project and the next summer he circulated a First Draft Report on the EDVAC. This set out the principles on which a fully effective general purpose computer, as the term is now understood, should be based. In particular it incorporated, for the first time, the concept of controlling the arithmetic and input/output operations via instructions that were stored in the memory, rather than held on punched tape or represented by cables and switches.

Such instructions could be fetched and decoded at a speed comparable with that of the arithmetic unit. Moreover, programs could be prepared away from the machine while it was busy on some other computation, and quickly loaded when required merely by being read into the machine’s storage unit.

It was only later, in 1947, that it was appreciated that this would allow instructions themselves to be generated, and perhaps changed, during the course of a computation. Address calculations could be performed in order to select specific data items from storage -

The very great majority of computers constructed for many years broadly used this von Neumann architecture.

However, even now, it is difficult to know just how great was the contribution John von Neumann (a brilliant mathematician but a great publicist) made to these important ideas, since so much originated from Mauchly and Eckert. This question of priority of ideas has remained a matter of bitter dispute for decades.

Von Neumann had met Turing (and Newman) prior to 1939, and was familiar with his work. However, though Turing influenced some of von Neumann’s work, it is unclear how far he may have influenced von Neumann's EDVAC report.

John von Neumann

The von Neumann Architecture