Abstract
GPU clusters have become attractive parallel platforms for high performance computing due to their ability to compute faster than the CPU clusters. We use this architecture to accelerate the mathematical operations of the GMRES method for solving large sparse linear systems. However the parallel sparse matrix-vector product of GMRES causes overheads in CPU/CPU and GPU/CPU communications when exchanging large shared vectors of unknowns between GPUs of the cluster. Since a sparse matrix-vector product does not often need all the unknowns of the vector, we propose to use data compression and decompression operations on the shared vectors, in order to exchange only the needed unknowns. In this paper we present a new parallel GMRES algorithm for GPU clusters, using compression vectors. Our experimental results show that the GMRES solver is more efficient when using the data compression technique on large shared vectors.
This work was supported by RƩgion de Franche-ComtƩ.
Chapter PDF
Similar content being viewed by others
References
Saad, Y., Schultz, M.: GMRES: a Generalized Minimal Residual Algorithm for Solving Nonsymmetric Linear Systems. SIAM J. Sci. Stat. Comput.Ā 7(3), 856ā869 (1986)
Jost, T., Contassot-Vivier, S., Vialle, S.: An Efficient Multi-algorithms Sparse Linear Solver for GPUs. In: EuroGPU Mini-Symposium of ParCo 2009, Lyon, pp. 546ā553 (2009)
Wang, M., Klie, H., Parashar, M., Sudan, H.: Solving Sparse Linear Systems on NVIDIA Tesla GPUs. In: Allen, G., Nabrzyski, J., Seidel, E., van Albada, G.D., Dongarra, J., Sloot, P.M.A. (eds.) ICCS 2009. LNCS, vol.Ā 5544, pp. 864ā873. Springer, Heidelberg (2009)
Ghaemian, N., Abdollahzadeh, A., Heinemann, Z., Harrer, A., Sharifi, M., Heinemann, G.: Accelerating the GMRES Iterative Linear Solver of an Oil Reservoir Simulator using the Multi-Processing Power of Compute Unified Device Architecture of Graphics Cards (2010)
Abbas-Turki, L., Vialle, S., Lapeyre, B., Mercier, P.: High Dimensional Pricing of Exotic European Contracts on a GPU Cluster, and Comparison to a CPU Custer. In: IPDPS 2009, pp. 1ā8. IEEE Computer Society (2009)
Bahi, J., Couturier, R., Ziane Khodja, L.: Parallel GMRES Implementation for Solving Sparse Linear Systems on GPU Clusters. In: HPC Symposium, pp. 23ā30. ACM/SIGSIM, Boston (2011)
Jeannot, E.: Improving Middleware Performance with AdOC: An Adaptive Online Compression Library for Data Transfer. In: IPDPS, vol.Ā 1, p. 70. IEEE, USA (2005)
Nvidia: NVIDIA CUDA C Programming Guide, Version 3.1.1 (2010)
Nvidia: Cuda Cublas Library, Version 3.1 (2010)
CUSP library, http://code.google.com/p/cusp-library/
Davis, T., Hu, Y.: The University of Florida Sparse Matrix Collection (1997), http://www.cise.ufl.edu/research/sparse/matrices/
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
Ā© 2012 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Bahi, J.M., Couturier, R., Khodja, L.Z. (2012). Parallel Sparse Linear Solver GMRES for GPU Clusters with Compression of Exchanged Data. In: Alexander, M., et al. Euro-Par 2011: Parallel Processing Workshops. Euro-Par 2011. Lecture Notes in Computer Science, vol 7155. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-29737-3_52
Download citation
DOI: https://doi.org/10.1007/978-3-642-29737-3_52
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-29736-6
Online ISBN: 978-3-642-29737-3
eBook Packages: Computer ScienceComputer Science (R0)