A GPU-based hyperbolic SVD algorithm

作者：Novakovic Vedran^*; Singer Sanja

来源：BIT Numerical Mathematics, 2011, 51(4): 1009-1030.

DOI：10.1007/s10543-011-0333-5

• One-sided Jacobi algorithm
• Hyperbolic singular value decomposition
• Symmetric indefinite eigenvalue problem
• GPU parallel programming

A one-sided Jacobi hyperbolic singular value decomposition (HSVD) algorithm, using a massively parallel graphics processing unit (GPU), is developed. The algorithm also serves as the final stage of solving a symmetric indefinite eigenvalue problem. Numerical testing demonstrates the gains in speed and accuracy over sequential and MPI-parallelized variants of similar Jacobi-type HSVD algorithms. Finally, possibilities of hybrid CPU-GPU parallelism are discussed.

• 出版日期2011-12