CPU Sequential Computing vs. GPU ParallelComputing: A Comparative Study for theSolution of Complex Problems

Abstract

Objective: To demonstrate the advantages of parallel programming using a graphics card for the solution of issues that would otherwise require excessive execution times.
Design/Methodology/Approach: A matrix inversion method was implemented for both sequential and parallel processes. The programmed processes were executed for different dimensions of the matrix to be inverted and the resulting execution times were compared.
Results: The reduction in processing time enables the inversion of large matrices in a short time (seconds); it can be scalable for other problems whose solution requires numerous operations and/or calculations.
Findings/Conclusions: The examples indicate the need for adequate computing equipment for highperformance tasks. Specifically, GPUs (graphics processing units) will make a greater contribution to the reduction of execution times than CPUs (central processing units).