论文信息 - Automatic Performance Tuning for the Multi-section with Multiple Eigenvalues Method for Symmetric Tridiagonal Eigenproblems

Automatic Performance Tuning for the Multi-section with Multiple Eigenvalues Method for Symmetric Tridiagonal Eigenproblems

We propose multisection for the multiple eigenvalues (MME) method for determining the eigenvalues of symmetric tridiagonal matrices. We also propose a method using runtime optimization, and show how to optimize its performance by dynamically selecting the implementation parameters. Performance results using a Hitachi SR8000 supercomputer with eight processors per node yield (1) up to 6.3x speedup over a conventional multisection method, and (2) up to 1.47x speedup over a statically optimized MME method.

[1] H. Simon. Bisection is not Optimal on Vector Processors , 1989 .

[2] Matteo Frigo,et al. A fast Fourier transform compiler , 1999, SIGP.

[3] Yuefan Deng,et al. New trends in high performance computing , 2001, Parallel Computing.

[4] E. Jason Riedy,et al. Benefits of IEEE-754 Features in Modern Symmetric Tridiagonal Eigensolvers , 2006, SIAM J. Sci. Comput..

[5] Ahmed H. Sameh,et al. A multiprocessor algorithm for the symmetric tridiagonal eigenvalue problem , 1985, PPSC.

[6] James Demmel,et al. Optimizing matrix multiply using PHiPAC: a portable, high-performance, ANSI C coding methodology , 1997, ICS '97.

[7] Jack J. Dongarra,et al. Automated empirical optimizations of software and the ATLAS project , 2001, Parallel Comput..

[8] B. Parlett,et al. Multiple representations to compute orthogonal eigenvectors of symmetric tridiagonal matrices , 2004 .

[9] Inderjit S. Dhillon,et al. The design and implementation of the MRRR algorithm , 2006, TOMS.

[10] Victor Eijkhout,et al. Self-Adapting Linear Algebra Algorithms and Software , 2005, Proceedings of the IEEE.