Developing High-Performance Parallel Applications Using EPAS

Abstract

In spite of the advent of high performance parallel computers and commodity clusters, complexity of parallel application development remains one of the major obstacles towards the mainstream adoption of parallel computing. Researchers are constantly investigating different approaches to reduce parallel application development time and increase productivity. As re-usable components, patterns have gained popularity in the sequential programming domain. Subsequently, several pattern-based parallel programming environments (PPEs) have been proposed to facilitate parallel application development procedure. Unfortunately, most of these PPEs lack the required flexibility in order to develop real-life parallel applications. In this paper, we describe the features of the EPAS (Extended Parallel Architectural Skeleton) PPE that enables development of complex parallel applications. We investigate and design the required patterns, and then use them to develop a parallel data cube computing application. Finally, we present the performance of the developed applications and discuss the results.