Training-based Workforce Development in Advanced Computing for Research and Education (Acore)

Semir Sarajlic,Neranjan Edirisinghe,Yubao Wu,Yi Jiang,Gregori Faroux

DOI: https://doi.org/10.1145/3093338.3104178

2017-01-01

Abstract:Data proliferation across all domains is enabled by the advancement in data sharing and procurement that has led to explosion in volume of data worldwide. The exponential growth in data has provided opportunities for machine learning algorithms and computational algorithms that benefit from greater availability of data. Access to robust and resilient high performance computing (HPC) resources is available through campus and/or national centers. The growth in demand for advanced computing resources from non-traditional HPC research community leads to an increasing strain on user-support at local centers and/or national programs such as XSEDE, which creates challenges. At Georgia State University (GSU), we have a diverse HPC user community of over 300 users whom represent more than 40 disciplines and use about four million CPU hours annually. We implemented a workforce development model to support our HPC community through a combination of local and external workshops as well as integration of HPC training within classes such as Big Data Programming, Scientific Computing, Parallel and Distributed Computing (PDC). As part of our approach, we consolidated our three disparate local HPC resources under one management system that provides a similar user environment to that of national resources part of XSEDE, which simplifies user transition from local to national resources.