University of Houston
Department of Computer Science
In partial fulfillment of the Requirements for the Degree of
Doctor of Philosophy
Zhenying Liu
will defend her dissertation proposal
Improving the Scalability of OpenMP Programs for Large Systems
Abstract
OpenMP has emerged as a parallel programming interface for shared memory
systems, and is supported on distributed shared memory systems (DSMs)
and some software distributed shared memory systems (Software DSMs).
OpenMP has the scalability problems for DSMs and software DSMs.
Major reasons are a lack of features for specifying data locality,
and a large amount of synchronization.
We propose to translate OpenMP programs into SPMD style to improve the data locality, and use a dynamic compilation approach to relax the synchronization. In the SPMD style, some shared arrays in OpenMP programs are privatized. Translation strategies are exploited for OpenMP constructs. On the other hand, global synchronization can be reduced after some dynamic information is known, such as the number of threads, and loop bounds. We will explore dynamic compilation techniques to generate efficient codes relying on feedback at runtime.
Date: Friday, Januaty 30, 2004
Time: 10 AM
Place: 550-PGH
Faculty, students, and the general public are invited.
Thesis Advisor: Dr. Barbara M. Chapman