The National Center for Supercomputing Applications (NCSA) will lead a team that has been awarded a three-year grant of $1.4 million from the National Science Foundation to develop a better way to manage the development of complex science and engineering codes.
As supercomputers become more powerful, they also become more complex. In order to take advantage of the increased power, scientific applications also will have to become more complex and will have to take advantage of more processing cores. Even those who are expert at optimizing these applications are quickly being overwhelmed.
In response, a team led by NCSA's Jay Alameda—with co-principal investigators Steven Brandt at Louisiana State University, Allen Maloney at the University of Oregon, and Marc Snir at the University of Illinois, and assisted by Greg Watson, the Eclipse Parallel Tools Platform project leader at IBM—is developing a Workbench for HPC Applications (W-HPC). The goal is to provide a comprehensive, powerful, easy-to-use development environment that will improve the way experts develop, debug, optimize, and run their science and engineering applications.
The team plans to build on the open-source Eclipse Parallel Tools Platform (Eclipse PTP), enhancing it to address completeness, scalability, debugging, integration, and specific platform issues, making it the preferred way to manage computational science and engineering code development, particularly for the world's largest computing systems. NCSA plans to use W-HPC on the Blue Waters supercomputer, an IBM POWER7 system that will tackle grand-challenge science and engineering problems.
The project also will encourage use and adoption of the W-HPC by providing documentation and training materials, online support, workshops and user group meetings, and college-level course materials.