Report 2009-13

Reduction to condensed forms for aymmetric eigenvalue problems on multi-core architectures

P. Bientinesi, F.D. Igual, D. Kressner, E.S. Quintana-Orti

Abstract: We investigate the performance of the routines in LAPACK and the Successive Band Reduction (SBR) toolbox for the reduction of a dense matrix to tridiagonal form, a crucial preprocessing stage in the solution of the symmetric eigenvalue problem. The target architecture is a current general purpose multi-core processor, where parallelism is extracted using a tuned multi-threaded implementation of BLAS. Also, in response to the advances of hardware accelerators, we modify the code in SBR to accelerate the computation by off-loading a significant part of the operations to a graphics processor (GPU). Our results on a system with two Intel QuadCore processors and a Tesla C1060 GPU illustrate the performance and scalability delivered by these architectures.

Paper: Available as PDF (242 KB) or as hardcopy to order reports@sam.math.ethz.ch.