Parallelization with MPI on 1-D Active Thermal Control Problem

Abstract

This paper describes a 1-D Active Thermal Control Problem (1-D ATCP) using parallel domain decomposition with parallel communication approach of Message Passing Interface (MPI) on Stationary Iterative Techniques (Jacobi and Gauss-Seidel. The discretization of the resulted matrices from the 1-D ATCP shows the ease of parallel implementation across the different domain of block sizes. The Single Program Multiple Data (SPMD) techniques was used to describe the parallelization of the method. We implemented the parallelization on Geo Cluster having to exploit inherent parallelism. Hence, the parallelization strategies are discussed, and results of the parallel experiments are presented with the speedup and efficiency on different block sizes agree with the theoretical analysis.