SPACE OF SIMD COMPUTERS

BY-Anissa Yagiek

CONTENTIntroductionSIMD computer organizationSpace of SIMD computersArray & associative processorsSIMD computer perspectiveApplications

Introduction SIMD is a class of parallel computers in Flynn's

taxonomy.such machines exploit  data level parallelism, but

not concurrency. SIMD is particularly applicable to common tasks

like adjusting the contrast in a digital image or adjusting the volume of digital audio.

Most modern CPU designs include SIMD instructions in order to improve the performance of multimedia use.

SIMD computer organizationIn SIMD instructions there are multiple

processing elements supervised by the same control unit.

All PEs receive the same instruction broadcast from the control unit but operate on different data sets from distinct data streams.

The shared memory subsystems may contain multiple modules.

Space of SIMD computersWe use the term array processor exclusively for

SIMD computers using conventional random-access memory and the term associative processor for SIMD computers using associative memory.

we divide the spaces of SIMD computers into five subspaces based on word-slice and bit-slice processing and number of control unit used

Word-slice array processorsBit-slice array processorsWord-slice associative processorsBit-slice associative processorsMultiple-SIMD computers

Multiple-SIMDMultiple-SIMD computers form a special

subclass of MIMD computersMultiple instruction streams exits in multiple-

array processorEach instruction stream handles multiple

data sets,as SIMD array

Array and associative processors

In 1958 Unger conceived a computer structure for spatial problems

A two-dimensional array of PEs is controlled by a common master

Unger’s machine was proposed for pattern –recognition applications

VAMPIn 1965 a vector arithmetic

multiprocessor(VAMP) was designed. It consists of a linear array of PEs with shared

memory modules and a shared arithmetic pipeline

each PE is a virtual processor, having only a few working registers in it

This pipeline-array processor was designed to save hardware in vector processing.

SIMD computer perspectivesSIMD computers are special-purpose systems.For a specific problem environment they may

perform impressively.however array processors have some

programming and vectorization problems which are difficult solve.

The reality is that array processor are not popular among commercial computer manufacturers

The performance is indicated under ideal programming and resource allocation conditions.

As the size of PE-array increases the performance should increase linearly.

The peak speed is also a function of the word length especially for bit-slice operations.

For vector processing the performance depends also on the vector length.

Application areas that have been suggested for array processors

Matrix algebraLinear and integer programmingGeneral circulation weather modelingImage processing and pattern recognitionWind-tuned experimentsAutomated map generationReal-time scene analysis