Performance Evaluation of Shake&wARP, A Cluster Enabled Application for Macromolecular Crystallography

The deployment of Shake&wARP, an application used in macromolecular crystallography for electron density map refinement, in a computational cluster managed by SUN's N1 Grid Engine presented and its performance is evaluated.In this deployment, parallelization was introduced in the optimization cycle, by decoupling the various optimization tasks. This implementation of Shake&wARP on a cluster improved the turnaround time of submitted jobs significantly. The performance of the implementation was evaluated using common UNIX utilities. Performance data gathered across the cluster were used to quantify the impact of the various software components of the clustered implementation of Shake&wARP on the cluster resources and to identify bottlenecks.