Optimization techniques for parallel processing

Techniques for optimization on a parallel architecture are presented, and are applied to a proprietary simulation code. Software tools available on most UNIX systems are used extensively in these techniques. Methods for determining where to partition an application for maximum benefit from parallelism are discussed, along with a way for determining what portion of the application has been parallelized. Finally, an algorithm is given which can be used to predict the effect on computation time of adding additional processors.