Understand the stored program concept, the Von Neumann architecture, the Harvard architecture, and the differences between them.

What this dot point is asking

AQA wants you to explain the stored program concept and describe the Von Neumann and Harvard architectures, including the differences between them.

The stored program concept

This idea, with the universal Turing machine as its theoretical basis, is what makes a general-purpose computer possible. Before stored programs, machines were configured for a single task by physical wiring; storing the program as data meant the same hardware could be repurposed instantly by loading new instructions. It is also why the fetch-decode-execute cycle works the way it does: the processor simply reads whatever instruction sits at the program counter, with no distinction in storage between a program and the numbers it manipulates.

The Von Neumann architecture

The bottleneck is the defining disadvantage and a favourite exam point. As processors became far faster than memory, the single shared path between processor and memory increasingly became the limiting factor, since every instruction fetch and every data access must take turns on the same bus. Techniques such as caching exist partly to soften this bottleneck by keeping frequently used instructions and data close to the processor.

The Harvard architecture

The hybrid approach used in most modern processors is worth noting: main memory follows the Von Neumann model (one store for everything), but the cache closest to the processor is split into separate instruction and data caches, giving Harvard-style simultaneous access where it matters most for speed. This shows that the two architectures are not rigid alternatives but design choices that real systems blend.