Toward exploitation of cell multi-processor array in time-consuming applications by using CNN model

  • Authors:

  • Zoltán Nagy;László Kék;Zoltán Kincses;András Kiss;Péter Szolgay

  • Affiliations:

  • Computer and Automation Research Institute, Hungarian Academy of Sciences, Budapest, Hungary;Computer and Automation Research Institute, Hungarian Academy of Sciences, Budapest, Hungary and Department of Image Processing and Neurocomputing, University of Pannonia, Veszprém, Hungary;Department of Information Technology, Pázmány Péter Catholic University, Budapest, Hungary;Department of Image Processing and Neurocomputing, University of Pannonia, Veszprém, Hungary;Computer and Automation Research Institute, Hungarian Academy of Sciences, Budapest, Hungary and Department of Image Processing and Neurocomputing, University of Pannonia, Veszprém, Hungary

  • Venue:
  • International Journal of Circuit Theory and Applications - Cellular Wave Computing Architecture
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Array computers can be useful in the solution of numerical spatio-temporal problems such as the state equation of the CNN or partial differential equations (PDE). IBM has recently introduced the cell broadband engine (cell BE) architecture, which contains eight identical vector processors in an array structure. In this paper the implementation of CNN simulation kernel on the cell BE is described. The simulation kernel is optimized according to the special requirements of the cell BE and can implement linear as well as nonlinear (piecewise linear) templates. In addition, the 2-D Princeton ocean model is implemented on the cell BE. The area-speed-power tradeoffs of our solution and different hardware implementations are also compared. Copyright © 2008 John Wiley & Sons, Ltd.