Parallel technologies for solving system of the linear equations by the conjugate gradient method
Dept. of Applied Mathematics, Oles Honchar Dnipropetrovs’k National University, Dnipropetrovs’k, Ukraine
Oles Honchar Dnipropetrovs’k National University, 2014
@article{bondarenko2014parallel,
title={Parallel technologies for solving system of the linear equations by the conjugate gradient method},
author={Bondarenko, Eduard},
year={2014}
}
The main purpose of this work is to show the clear advantages of using modern parallel tools in solving the SLAE. The BiCGStab method was used for solving system of the linear equations. This paper contains some details about this method. To accelerate computations on the GPU several technologies (such as CUBLAS, OpenACC) were used. The main novelty of this paper is the comparing the method (BiCGStab) from the AMGX library with the own implementation of this method. How you can see the AMGX library provides not very optimal working time and memory usage. The part of this paper contains some assumptions about the results of this library. At the end this paper contains some other ways of obtaining the performance gain. Also this paper includes some thoughts about the usage the OpenMP technology in this type of the calculations.
June 14, 2014 by hgpu
