💻

Computer Architecture Classifications

Jul 26, 2025

Overview

This lecture explains the two most widely used classifications of computer architecture: the Von Neumann vs. Harvard models, and Flynn's taxonomy based on instruction and data streams.

Classification by Architecture Models

  • Computer architecture is commonly classified as either Von Neumann or non-Von Neumann models.
  • Von Neumann architecture contains three subsystems: CPU, main memory, and I/O system.
  • It is a stored program computer, meaning both programs and data are kept in main memory.
  • Instructions are executed sequentially, with only one operation handled at a time.
  • The Von Neumann bottleneck arises due to a single path for both data and instructions between memory and CPU.
  • Harvard architecture uses separate memory units for instructions and data, enabling parallel access and faster performance.
  • Modified Harvard architecture introduces a cache that allows the processor to act as a Harvard or Von Neumann machine, depending on memory access.

Flynn’s Taxonomy

  • Flynn's taxonomy classifies computers by the number of instruction and data streams.
  • SISD (Single Instruction stream, Single Data stream): One CPU, one instruction at a time, one data item at a time—typical Von Neumann machines.
  • SIMD (Single Instruction stream, Multiple Data streams): One control unit; multiple ALUs perform the same instruction on different data items simultaneously.
  • MISD (Multiple Instruction streams, Single Data stream): Hypothetical class; different instructions operate on the same data, rarely implemented in practice.
  • MIMD (Multiple Instruction streams, Multiple Data streams): Multiprocessor systems where each processor executes independent instructions on separate data—common in modern systems.

Key Terms & Definitions

  • Von Neumann Architecture — Computer model with shared memory for program instructions and data, sequential instruction execution.
  • Harvard Architecture — Computer model using separate memories and pathways for instructions and data, enabling parallel access.
  • Von Neumann Bottleneck — Limitation caused by a single memory pathway for both data and instructions.
  • Flynn’s Taxonomy — A system for classifying computer architectures by instruction and data stream counts (SISD, SIMD, MISD, MIMD).
  • SISD — Single CPU, one instruction, one data path.
  • SIMD — Single instruction, multiple data units processed in parallel.
  • MISD — Multiple instructions, single data stream, rare in practice.
  • MIMD — Multiple instructions and data streams; multiprocessor systems.
  • Cache — Small, fast memory storage used to speed up access; central to modified Harvard architectures.

Action Items / Next Steps

  • Review and compare the characteristics of Von Neumann, Harvard, and modified Harvard architectures.
  • Study examples of systems using SISD, SIMD, and MIMD for deeper understanding.
  • Prepare questions for next class discussion on computer architecture classifications.