Performance evaluation for a loosely coupled parallel processing environment

The advantages and limitations of distributed shared memory (DSM) in a loosely coupled parallel processing environment are examined. The method of examination is to evaluate the performance of parallel algorithms. The approach was to adapt sequential applications for a simulated parallel processing environment and at some future time be able to use the applications in the target DSM environment where actual parallelism is possible. In order to examine the issues used in parallel performance evaluation, timings and pertinent information were recorded. The information was assimilated and evaluated through the use of metrics. The results are used to evaluate the performance of the applications which are to be used in the loosely coupled parallel processing environment.<
>