A GaBP-GPU Algorithm of Solving Large-Scale Sparse Linear Systems
School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China
Journal of Information and Computational Science, Vol. 11 (3): 911-921, 2014
According to GaBP (Gaussian Belief Propagation) algorithm, this article presents a GaBP-GPU algorithm of solving large-scale symmetric diagonally dominant sparse linear systems based on GPU. Combined with GaBP-GPU algorithm, a storage format (MCSC) is presented. We extract some diagonally dominant matrices from the University of Florida Sparse Matrix Collection as test examples. The experimental results show that our algorithm has high efficiency in execution time under the same accuracy.
February 25, 2014 by hgpu
