The first ideas for a parallel computer architecture with maximum computing power, which could replace the dominant, sequentially operating Von Neumann architecture, date back to 1997. The rapid development and increasing use of multi-core processors retrospectively confirm the concept at that time, also under the aspect that the clock speed of the processors has not been increased for years. The parallel concept is developed at the University of Maryland, is called Explicit Multi-Threading (XMT) and is inseparably linked with the name of Professor Uzi Vishkin.
Unlike the Von Neumann computer, where individual instructions are processed serially, the XMT concept allows any number of instructions to be entered and processed simultaneously. This is achieved with Immediate Concurrent Execution (ICE), a programming support for the algorithms of the PRAMs and the XMT hardware, which ensures that the sequence of instructions makes sense.
In current multi-core systems, the operating system assigns work to the individual processors. The XMT concept is different, where the hardware distributes the work and one processor of the multi-core processors controls the other processors. So in this concept, the hardware has control over the load distribution. So that the data flow is not unnecessarily impaired by bus systems, the XMT concept provides for the processors and the main memories with their controllers to be located on the processor chip.