DocumentCode
2037098
Title
Massive computation of singular values of small matrices on GPUs
Author
Capozzoli, Amedeo ; Curcio, Claudio ; Di Donna, Salvatore ; Liseno, Angelo
Author_Institution
Dipt. di Ing. Elettr. e delle Tecnol. dell´Inf., Univ. di Napoli Federico II, Naples, Italy
fYear
2015
fDate
1-4 July 2015
Firstpage
1
Lastpage
2
Abstract
This paper presents a fast and robust approach to evaluate the singular values of a very large number of small matrices on GPUs. The timings and the accuracy obtained are compared to those achieved by "state-of-the-art" libraries. The results show that, by suitably tuning the trade-off between computational speed and accuracy, a reasonable speedup over a fast multi-core CPU is possible.
Keywords
graphics processing units; mathematics computing; singular value decomposition; GPU; computational speed; multicore CPU; singular value computation; small matrices; state-of-the-art libraries; Accuracy; Eigenvalues and eigenfunctions; Graphics processing units; Instruction sets; Jacobian matrices; Matrix decomposition; Symmetric matrices;
fLanguage
English
Publisher
ieee
Conference_Titel
Computational Electromagnetics International Workshop (CEM), 2015
Conference_Location
Izmir
Print_ISBN
978-1-4673-9279-2
Type
conf
DOI
10.1109/CEM.2015.7237428
Filename
7237428
Link To Document