Symbolic Analysis for Parallelizing Compilers

Abstract

High performance computers (HPCs) are of paramount importance in science and engineering. They are widely used in large-scale applications that simulate complex processes. Examples of such applications include weather modeling and forecasting, crash-tests, optimizing combustion engines, aerodynamic behavior of airplanes, study of chemical phenomena, and financial modeling [8], [39], [72], [13], [38]. Currently, HPCs offer a peak performance of several hundred TFlops and memory capacities of several TBytes. Even new systems achieving PFlops are in development. The two major sources contributing to the power of HPCs are (1) powerful hardware technology and (2) parallelism. Due to physical limits of hardware technology, parallelism plays a more important role in meeting the ever increasing demand of computational power.