The term "von Neumann architecture" arose from mathematician John von Neumann's paper, First Draft of a Report on the EDVAC. Dated June 30, 1945, it was an early written account of a general purpose stored-program computing machine (the EDVAC). However, while von Neumann's work was pioneering, the term von Neumann architecture does somewhat of an injustice to von Neumann's collaborators, contemporaries, and predecessors.
A patent application of Konrad Zuse mentioned this concept in 1936.
The idea of a stored-program computer existed at the Moore School of Electrical Engineering at the University of Pennsylvania before von Neumann even knew of the ENIAC's existence. The exact person who originated the idea there is unknown.
John William Mauchly and J. Presper Eckert wrote about the stored-program concept in December 1943 during their work on ENIAC. Additionally, ENIAC project administrator Grist Brainerd's December 1943 progress report for the first period of the ENIAC's development implictly proposed the stored program concept (while simultaneously rejecting its implementation in the ENIAC) by stating that "in order to have the simplest project and not to complicate matters" the ENIAC would be constructed without any "automatic regulation."
When the ENIAC was being designed, it was clear that reading instructions from punched cards or paper tape would not be fast enough, since the ENIAC was designed to execute instructions at a much higher rate. The ENIAC's program was thus wired into the design, and it had to be rewired for each new problem. It was clear that a better system was needed. The initial report on the proposed EDVAC was written during the time the ENIAC was being built, and contained the idea of the stored program, where instructions were stored in high-speed memory, so they could be quickly accessed for execution.
von Neumann bottleneck
The separation between the CPU and memory leads to the von Neumann bottleneck, the limited throughput (data transfer rate) between the CPU and memory compared to the amount of memory. In modern machines, throughput is much smaller than the rate at which the CPU can work. This seriously limits the effective processing speed when the CPU is required to perform minimal processing on large amounts of data. The CPU is continuously forced to wait for vital data to be transferred to or from memory. As CPU speed and memory size have increased much faster than the throughput between them, the bottleneck has become more of a problem.
- "Surely there must be a less primitive way of making big changes in the store than by pushing vast numbers of words back and forth through the von Neumann bottleneck. Not only is this tube a literal bottleneck for the data traffic of a problem, but, more importantly, it is an intellectual bottleneck that has kept us tied to word-at-a-time thinking instead of encouraging us to think in terms of the larger conceptual units of the task at hand. Thus programming is basically planning and detailing the enormous traffic of words through the von Neumann bottleneck, and much of that traffic concerns not significant data itself, but where to find it."
The performance problem is reduced by a cache between CPU and main memory, and by the development of branch prediction algorithms. It is less clear whether the intellectual bottleneck that Backus criticized has changed much since 1977. Backus's proposed solution has not had a major influence. Modern functional programming and object-oriented programming are much less geared towards pushing vast numbers of words back and forth than earlier languages like Fortran, but internally, that is still what computers spend much of their time doing.