Computer Architecture – Overview

This article is a term definition for computer architecture – including exam formula, key concepts, and tags.

In a Nutshell

The Von Neumann principle is the foundational model of modern computers: the CPU fetches instructions and data from a shared memory via a shared bus (potential bottleneck).

The Von Neumann Principle (Foundational Model)

Main components:

1. Arithmetic Logic Unit (ALU)
2. Control Unit
3. Memory for program and data
4. Input/Output (I/O)

All components are coupled via a shared bus. This creates the Von Neumann bottleneck, because instructions and data use “the same path”.

CPU – the “Brain”

Typical sequence: Fetch → Decode → Execute → Writeback.

Important concepts:

• Registers (fast, small, in the CPU)
• Instruction set (e.g., x86/ARM); CISC vs. RISC
• Clock frequency (GHz) is not the only performance metric
• Multi-core enables parallel processing

Memory Hierarchy

From fast/expensive to slow/cheap:

1. Registers
2. Cache (L1/L2/L3)
3. RAM
4. SSD/Hard Drive

Cache exploits the principle of locality:

• spatial locality
• temporal locality

Peripherals & Bus Systems

• Bus: data bus, address bus, control bus

I/O methods:

• Programmed I/O (Polling)
• Interrupt-driven I/O
• DMA (Direct Memory Access): data transfers directly between device and RAM without constant CPU involvement

Performance Evaluation

Performance depends not only on GHz, but also on:

• IPC (Instructions per Cycle)
• Number of cores
• Memory latency/bandwidth

Exam Formula (Oral)

A computer is a machine that operates according to the Von Neumann principle.
A CPU (ALU + control unit) fetches instructions and data from a shared memory
via a bus, performs calculations, and writes results back. To compensate for
slow memory access, there is a memory hierarchy with fast caches. I/O is
organized via interrupts or DMA to reduce CPU overhead.