Definition
A general-purpose computer architecture that stores both instructions and data in readable and writable memory and executes instruction sequences automatically.
What it is
A stored-program computer also treats instructions as addressable content held in memory. The processor uses the program counter to fetch, decode, and execute an instruction, then proceeds to the next instruction or a conditional branch.
Why it emerged and what it solved
Early machines often defined tasks through plugboards, switches, or fixed control logic, making changes of task costly. Once instructions were stored in memory, the computing process could be changed by modifying the program, turning the equipment from a special-purpose calculator into a general-purpose computer.
Why it mattered and what it enabled
It established the basic division of work between hardware and software, and made "faster, larger, more reliable memory" a central technological requirement. Delay-line, magnetic-core, and semiconductor memories all continued to evolve in response to this demand.
Principles and impact
Problems solved
- Avoided rewiring or manually configuring an entire set of control circuits every time the task changed.
- Made programs into data that could be saved, modified, and executed repeatedly.
Principles
- The processor fetches instructions from memory, decodes and executes them, and updates its state.
- Instructions and data use a unified addressable memory space or related memory structures.
Improvements
- Moved computational control from fixed wiring to reprogrammable software sequences.
- Allowed the same machine to perform different tasks by changing its program.
Significance
- Established the software-programmable foundation of modern general-purpose computers.
- Created continuing demand for memory capacity, speed, and random access.
History and applications
People
Tom Kilburn、Frederic C. Williams、John von Neumann
Organizations
University of Manchester、Institute for Advanced Study
Events
Manchester Baby runs a stored program in 1948、Manchester Mark 1 enters operation in 1951
Applications
Scientific computing、Cryptanalysis and engineering calculations、General-purpose commercial data processing
Limitations
- Early memory had small capacity and low speed, making instruction fetching and data access bottlenecks.
- Sharing paths between instructions and data creates bandwidth and security-isolation problems.
Upstream and downstream relations
Upstream
Sources: First Steps: Lectures from the Dawn of Computing1946: Williams demonstrates CRT storage
Downstream
Sources: A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence