Parallel sparse linear solver GMRES for GPU clusters with compression of exchanged data

Authors:
Jacques M. Bahi;Raphaël Couturier;Lilia Ziane Khodja
Affiliations:
LIFC laboratory, University of Franche-Comte, Belfort Cedex, France;LIFC laboratory, University of Franche-Comte, Belfort Cedex, France;LIFC laboratory, University of Franche-Comte, Belfort Cedex, France
Venue:
Euro-Par'11 Proceedings of the 2011 international conference on Parallel Processing
Year:
2011

Citing 6
Cited 0

GMRES: a generalized minimal residual algorithm for solving nonsymmetric linear systems

SIAM Journal on Scientific and Statistical Computing
Improving Middleware Performance with AdOC: An Adaptive Online Compression Library for Data Transfer

IPDPS '05 Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Papers - Volume 01
Solving Sparse Linear Systems on NVIDIA Tesla GPUs

ICCS '09 Proceedings of the 9th International Conference on Computational Science: Part I
High dimensional pricing of exotic European contracts on a GPU Cluster, and comparison to a CPU cluster

IPDPS '09 Proceedings of the 2009 IEEE International Symposium on Parallel&Distributed Processing
Parallel GMRES implementation for solving sparse linear systems on GPU clusters

Proceedings of the 19th High Performance Computing Symposia
The university of Florida sparse matrix collection

ACM Transactions on Mathematical Software (TOMS)

Quantified Score

Hi-index	0.00

Visualization

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.