Accelerating the Lyapack library using GPUs

Lyapack is a package for the solution of large-scale sparse problems arising in control theory. The package has a modular design, and is implemented as a Matlab toolbox, which renders it easy to utilize, modify and extend with new functionality. However, in general, the use of Matlab in combination with a general-purpose multi-core architecture (CPU) offers limited performance when tackling the sparse linear algebra operations underlying the numerical methods involved in control theory.In this paper we extend Lyapack to leverage the computational power of graphics processors (GPUs). The experimental evaluation of a new CUDA-enabled solver for the Lyapunov equation, a crucial operation appearing in control theory problems, shows a significant runtime reduction when compared with the original CPU version of Lyapack, while retaining the usability of a Matlab-based implementation.

[1]  Enrique S. Quintana-Ortí,et al.  A mixed-precision algorithm for the solution of Lyapunov equations on hybrid CPU-GPU platforms , 2009 .

[2]  BennerPeter,et al.  A mixed-precision algorithm for the solution of Lyapunov equations on hybrid CPU-GPU platforms , 2011, ParCo 2011.

[3]  Athanasios C. Antoulas,et al.  Approximation of Large-Scale Dynamical Systems (Advances in Design and Control) (Advances in Design and Control) , 2005 .

[4]  Athanasios C. Antoulas,et al.  Approximation of Large-Scale Dynamical Systems , 2005, Advances in Design and Control.

[5]  Dimitri Jeltsema,et al.  Proceedings of the European Control Conference, ECC 2003 , 2003 .

[6]  Pablo Ezzatti,et al.  A Study on the Implementation of Tridiagonal Systems Solvers Using a GPU , 2011, 2011 30th International Conference of the Chilean Computer Science Society.

[7]  Paul Van Dooren,et al.  A collection of benchmark examples for model reduction of linear time invariant dynamical systems. , 2002 .

[8]  V. Sundarapandian,et al.  Computational Methods for Linear Control Systems , 2011 .

[9]  Thilo Penzl,et al.  A Cyclic Low-Rank Smith Method for Large Sparse Lyapunov Equations , 1998, SIAM J. Sci. Comput..

[10]  Enrique S. Quintana-Ortí,et al.  Accelerating Model Reduction of Large Linear Systems with Graphics Processors , 2010, PARA.

[11]  Maxim Naumov,et al.  Incomplete-LU and Cholesky Preconditioned Iterative Methods Using CUSPARSE and CUBLAS , 2012 .

[12]  Serkan Gugercin,et al.  A survey of balancing methods for model reduction , 2003, 2003 European Control Conference (ECC).

[13]  Roger W. Hockney,et al.  A Fast Direct Solution of Poisson's Equation Using Fourier Analysis , 1965, JACM.

[14]  Mihail M. Konstantinov,et al.  Computational methods for linear control systems , 1991 .