Abstract :
— Explaining frameworks of linear equations are
presumably a standout amongst the most investigative
applications of linear algebra. Direct techniques for
registering the arrangement of such frameworks can be
exceptionally extravagant because of high memory
necessities and computational expense. This is a decent
motivation to utilize iterative techniques which processes
just an estimate of the arrangement. This paper overhauls
one of the iterative strategies for direct frameworks of
linear systems of equation on a cuda platform and openmp
platform. It analyzes how quick the parallel code runs than
that of serial code furthermore looks at the cpu and gpu
based parallel computing.
Keyword :
Graphics processing unit, Compute unified development architecture, Parallelization, Congugate gradient