Author(s)

G. Venter & B. Watson

Abstract

Exploiting parallelism in general purpose optimization G. Venter & B. Watson Vanderplaats Research and Development, Inc. Colorado Springs, CO, USA Abstract VisualDOC, a general purpose design optimization code that is commercially available from Vanderplaats Research and Development, Inc. (VR&D), is used as a test-bed for evaluating the efficiency of parallelizing current optimization algorithms. The finite difference gradient calculations of VisualDOC are implemented in parallel form and these changes are evaluated using a typical aircraft wing example problem. VisualDOC provides three algorithms for performing constrained nonlinear optimization and the investigation focuses on the influence of the optimization algorithm and the number of design variables on the efficiency of performing the optimization in parallel. The need to additionally parallelize the one-dimensional search calculations is also investigated. The Local Area Multiprocessor (LAM

Keywords