Flynn’s Taxonomyof Computer Architectures

Source: Wikipedia

Michael Flynn 1966

CMPS 5433 – Parallel Processing

Flynn’s Taxonomy

Proposed in 1966

General 4 category system

Does not clearly classify all models in use today

There are combinations, also.

Four Categories - Terminology

S = Single

I = Instruction Stream

M = Multiple

D = Data Stream

SISD

SIMD

MISD

MIMD

SISD

Single Instruction, Single Data stream A sequential computer which exploits no parallelism in either the instruction or data streams. Examples of SISD architecture are the traditional single processor machines like a PC (currently manufactured PC's have multiple processors) or old mainframes.

SISD

SIMDSingle Instruction, Multiple Data streamsExploits multiple data streams against a single instruction stream to perform operations which may be naturally parallelized.

For example, an array processor or GPU.

Typical for splitting large data sets.

SIMD

NOTE: Front-end processor

ONE processor starts the program– It sends code to the other processors

In diagrams, the front-end is represented by

which sends code to individual PU’s

which execute the remaining code.

Instruction Pool

PU

SIMD Code

Each processor executes the same code

Code may be variable based on processor id

Example: Sum elements of array A

for each Pi, i = 0 to 9 // “executed” by front-end

Ti = 0

for j = 0 to 9 // Pi sums its segment

Ti = Ti + A[i * 10 + j]

MISDMultiple Instruction, Single Data streamMultiple instructions operate on a single data stream. Uncommon architecture which is generally used for fault tolerance.

Systolic Array: process data & pass on to next PU

Examples include the Space Shuttle flight control computer.

Least common

MISD

MIMD

Multiple Instruction, Multiple Data streamsMultiple autonomous processors simultaneously executing different instructions on different data.

May send results to central location.

Distributed systems – can be shared memory space or distributed memory space.

MIMD