Accelerating the reduction to upper Hessenberg, tridiagonal, and bidiagonal forms through hybrid GPU-based computing
University of Tennessee (USA), Department of Electrical Engineering and Computer Science, 1122 Volunteer Blvd, Knoxville TN 37996-3450
Parallel Computing, Volume 36, Issue 12, December 2010, Pages 645-654 (30 June 2010)
@article{tomov2010accelerating,
title={Accelerating the reduction to upper Hessenberg, tridiagonal, and bidiagonal forms through hybrid GPU-based computing},
author={Tomov, S. and Nath, R. and Dongarra, J.},
journal={Parallel Computing},
issn={0167-8191},
year={2010},
publisher={Elsevier}
}
We present a Hessenberg reduction (HR) algorithm for hybrid systems of homogeneous multicore with GPU accelerators that can exceed 25 ? the performance of the corresponding LAPACK algorithm running on current homogeneous multicores. This enormous acceleration is due to proper matching of algorithmic requirements to architectural strengths of the system
November 17, 2010 by hgpu
