Solution of linear equations and performance analysis on desktop computing systems

In this paper we present an evaluation of two common algorithms for solving linear system of equations. The Gauss-elimination, Gauss-Jordan and Gauss-Jordan with tri diagonal elements algorithms were selected for the analysis and the performance evaluation is done on commonly used desk top systems with different processor speeds. The algorithms were implemented using the efficient floating point assembly language subroutines. The programs were executed on a variety of systems with single core and multi core processors and their performances were compared. The results show that the Gauss Jordan method is faster than the Gauss Elimination method. It is observed that the performance on core2duo systems is better than single core systems. The analysis also reveals that the speedup is much higher in systems populated with 64 bit processors.