Digital Sherpa

Abstract

Currently users of high performance computers are overwhelmed with non scalable tasks such as job submission and monitoring, a problem that gets compounded when trying to run complex scientific applications requiring the coordination of several interrelated programs. Digital Sherpa (DS) is a grid tool set for coordinating the execution of multiple jobs on separate HPC resources; DS automates non-scalable tasks such as job submission and monitoring, and includes recovery features such as resubmission of failed jobs and program restarting. DS has been used to develop a Grid enabled version, MGAC-CGA, of the Modified Genetic Algorithms for Crystals and Clusters (MGAC), a parallel distributed application for the prediction of the structures of atomic clusters and organic crystals using Genetic Algorithms (GA). MGAC-CGA has been successfully tested on the NSF TeraGrid and on several clusters at the University of Utah.