2953

Iterative GPGPU Linear Solvers for Sparse Matrices

Filip Vesely
Faculty of Electrical Engineering, Czech technical University in Prague
Faculty of Electrical Engineering, Czech technical University in Prague, 2008

@article{vesely2008iterative,

   title={Iterative GPGPU Linear Solvers for Sparse Matrices},

   author={Vesel{‘y}, B.F.},

   year={2008}

}

Download Download (PDF)   View View   Source Source   

830

views

The performance and the level of programmability of graphics processors (GPU) on current video cards offer new capabilities beyond the graphics applications for which they were designed. These are general-purpose computations which expose parallelism. In this thesis, I describe the iterative methods for solving sparse linear systems: the Jacobi, Gauss-Seidel, Conjugate Gradient and BiConjugate Gradient method. I explain the architecture of the GPU, their capability of the general-purpose computations and current software platforms which provide an abstraction to the GPU programming: the BrookGPU and RapidMind Multi-core platform. I discuss the iterative methods in the GPU programming environment and describe the implementation for both platforms. Then I made a performance tests between the sequential algorithms and the algorithms for Brook and Rapidmind. The results of my experiments show that the GPU is better suited for the computation with larger problem dimensions. For the iterative methods, the optimal problem size ~ 10^6 was observed, where most significant speedup compared to the sequential implementation was reached. Namely 3x speedup for Brook and 1.5x speedup for Rapidmind. Larger instances (10^7+) encountered a limitation of the graphics API. Brook has been found as faster platform than Rapidmind due to the necessary workarounds in Rapidmind implementation. The general Gauss-Seidel method has been found as not suitable for processing on the GPUs.
No votes yet.
Please wait...

* * *

* * *

Featured events

2018
November
27-30
Hida Takayama, Japan

The Third International Workshop on GPU Computing and AI (GCA), 2018

2018
September
19-21
Nagoya University, Japan

The 5th International Conference on Power and Energy Systems Engineering (CPESE), 2018

2018
September
22-24
MediaCityUK, Salford Quays, Greater Manchester, England

The 10th International Conference on Information Management and Engineering (ICIME), 2018

2018
August
21-23
No. 1037, Luoyu Road, Hongshan District, Wuhan, China

The 4th International Conference on Control Science and Systems Engineering (ICCSSE), 2018

2018
October
29-31
Nanyang Executive Centre in Nanyang Technological University, Singapore

The 2018 International Conference on Cloud Computing and Internet of Things (CCIOT’18), 2018

HGPU group © 2010-2018 hgpu.org

All rights belong to the respective authors

Contact us: