The overall objective of this research was to develop instructional modules demonstrating the use of performance software in programming numerical algorithms and their implementations on concurrent processor systems.

Developing correct and efficient software is far more complex for parallel systems than it is for sequential processors.

As part of this development and implementation, performance software (such as VAMPIR, PARAVER, etc.) were used to help analyse and understand the execution behaviour. The goal was to identify program regions that do not perform well. With these, communication patterns, processor load balance, computation versus comunication ratios, timing characteristics, and processor idle times were obtained and studied. Furthermore, an evaluation of these performance software tools were made with the goal of testing their effectiveness and developing extensions to them.

Acknowledgment and Disclaimer: This material is based upon work supported by the National Science Foundation under Grant No. 0086262. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.