Implementations and Performance of Nonlinear CG Methods by TAO on Dawning2000

  • Authors:

  • Jian Wang;Xuebin Chi;Tongxiang Gu

  • Affiliations:

  • Chinese Academy of Sciences, Beijing, P.R. China;Chinese Academy of Sciences, Beijing, P.R. China;Institute of Applied Physics and Computational Mathematics, Beijing, P.R. China

  • Venue:
  • HPCASIA '04 Proceedings of the High Performance Computing and Grid in Asia Pacific Region, Seventh International Conference
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Nonlinear conjugate gradient methods (CG) are typical unconstrained optimization methods. As the optimization problems to be solved become larger, the dependence on efficient and scalable software is severe. Toolkit for Advanced Optimization (TAO) is a parallel package that can currently solve several kinds of optimization problems. In this paper, we give the framework of several variants of CG: CG_FR, CG_PR, CG_PRP and their implementations in TAO1.5, which have been tested up to 64 processors on Dawning2000 to solve problems with up to 10驴 variables. The results show that the scalability of CG implementations in TAO1.5 is excellent.