Experiments in parallel processing for undergraduate students

Abstract

A series of experiments were designed to allow students to gain experience with several problems associated with parallel programming. A prime number generator was used for each of the experiments. In the first experiment, the prime number generator was run on a single processor. The results were used as a baseline for each of the following experiments. The second experiment given to the students had them modify the prime number generator to run on a set of parallel processors. The students expected the program's run time to decrease by a factor of the number of processors. However, because of the communication method chosen, the run time only decreased by a minimal amount. The students were required to analyze why this happened. The next experiment tried to resolve some of the problems uncovered in the previous experiment. New problems were introduced that produced deadlock. The students used graph theory techniques to show why deadlocked occurred. The last experiment developed was a message passing system. The solution was run on several communication configurations including linear and hypercube [De90, Ka87]. This allowed the students to see how the structure of the parallel communication structure impacted the efficiency of the solution. The experiments were designed to allow the students to gain experience with parallel processing and some of the problems associated with it. The experiments were successful in that the obvious solution did not work as planned. The students were required to analyze why the solutions did not work and to devise alternative solutions.