Sparse approximate inverse preconditioners for iterative solvers on GPUs

Authors:
Mykola Lukash;Karl Rupp;Siegfried Selberherr
Affiliations:
Institute for Microelectronics, TU Wien, Wien, Austria;Institute for Microelectronics, TU Wien, Wien, Austria and Institute for Analysis and Scientific Computing, TU Wien Wiedner, Wien, Austria;Institute for Microelectronics, TU Wien, Wien, Austria
Venue:
Proceedings of the 2012 Symposium on High Performance Computing
Year:
2012

Citing 7
Cited 0

BI-CGSTAB: a fast and smoothly converging variant of BI-CG for the solution of nonsymmetric linear systems

SIAM Journal on Scientific and Statistical Computing
Parallel Preconditioning with Sparse Approximate Inverses

SIAM Journal on Scientific Computing
Multigrid

Multigrid
Factorized Sparse Approximate Inverses for Preconditioning

The Journal of Supercomputing
Iterative Methods for Sparse Linear Systems

Iterative Methods for Sparse Linear Systems
FSAI preconditioned CG algorithm combined with GPU technique for the finite element analysis of electromagnetic scattering problems

Finite Elements in Analysis and Design
A parallel algebraic multigrid solver on graphics processing units

HPCA'09 Proceedings of the Second international conference on High Performance Computing and Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

Iterative solvers with preconditioners are typically employed for the solution of large systems of linear equations. However, the design of preconditioners for the black-box case, in which no additional information about the underlying problem is known, is very difficult. The most commonly employed method of incomplete LU factorizations is a serial algorithm and thus not well suited for the massively parallel computing architecture of GPUs. We investigate sparse approximate inverse preconditioners in this work, which show a very high degree of parallelism. The preconditioner setup is accomplished in a hybrid manner, where parts of the algorithm which require dynamic memory allocations are carried out on the CPU, while the GPU is used for the computationally expensive factorizations. Our benchmark results demonstrate that our implementations in ViennaCL are well suited as a black-box preconditioner for multi- and many-core architectures.