University of Houston
Department of Computer Science
In
partial fulfillment of the Requirements for the Degree of
Master of
Science
Xiaoyu Tan
will defend her
thesis
Parallel Reflectivity Method On Beowulf Clusters
Abstract
In this project, we implemented the reflectivity method, which was introduced in seismology by Thomson, using the object oriented programming language C++. The program was then parallelized using the Message Passing Interface (MPI). The parallelized version focused on parallel execution on the newly installed Beowulf cluster system at the University of Houston, Allied Geophysical Laboratories (AGL). After incorporating MPI into my sequential implementation, the parallelized version of the program carries out the reflectivity computation on one, several, or all processors of the cluster system simultaneously. The developed program has been shown to produce correct results for certain cases when the mathematical solution can be computed by hand. The parallelized version significantly reduces execution time when multiple processors are used, from approximately 49 percent (two processors) up to 95 percent (23 processors) compared to running on a single processor. We concentrate the discussion on the design and implementation of the reflectivity method on the Beowulf cluster system.
Date: Tuesday, October 30, 2001
Time: 11:30 AM
Place:
550-PGH
Faculty, students, and the general public are
invited.
Thesis Advisor: Dr. Olin Johnson