Parellel processing of arithmetic expressions

Abstract

As the need for the fast Computer system increases, the use of parallel processing in computer system becomes important. This thesis deals with the parallel processing of arithmetic expressions in high level language programs. In this thesis we propose a multiprocessor architecture which can efficiently execute arithmetic expressions. Parallelism detection is performed by using an algorithm based on Paralled Execution String for detecting parallelism in the program block level. In order to apply this algorithm, the arithmetic expressions must first be converted into the reverse polish strings then Parallel Execution Strings must be obtained from them. A multiprocessor system is designed for parallel processing of the parallel execution strings. Simulation results show that it executes the sample programs 1.5-2.5 times faster than the uniprocessor system.