Michael Levine and Ralph Roskies

Foreword from the Directors, 2005

In this booklet we highlight the past year’s scientific progress at PSC. As with half a dozen prior state-of-the-art systems, PSC has demonstrated its ability to rapidly transform new technologies into productive research tools. The first XT3 system shipped from Cray, Big Ben (Creating Cyberinfrastructure) has in the space of a year been installed, tested, optimized for research applications and is now a production resource on the TeraGrid. Such feats happen only because PSC’s exceptional systems and networking staff (Networking the Future) possess the world-class knowledge, experience, creativity and energy to make them happen.

With Sandia National Laboratories and Cray Inc., PSC collaborated to envision an architecture, realized in the XT3, that represents the most advanced current thinking for a high-performance system to support very large-scale computations across many fields, and that is fully expandable to well over 100 teraflops. With PSC’s active leadership, Big Ben became the first Cray XT3 to run applications, first to be made available to NSF researchers, and has already produced notable results in several fields, including turbulence ( Big Ben Delivers in Real Time via TeraGrid) and nanotechnology (Dwarf Materials).

This entirely new architecture is now integrated into the TeraGrid, where it will become the resource best suited for very large-scale, demanding projects. The United States needs leading-edge capability to continue to advance in science and engineering, and we are pleased that the NSF has decided to push to a Petaflop computer by 2010.

The TeraGrid itself, NSF’s visionary program to transform the nation’s research capability, is now entering a second major phase, with PSC in a leadership role — supporting the research community in exploiting the TeraGrid as a resource to produce new knowledge, as we have done with PSC systems, and leading the TeraGrid’s increasingly vital security efforts. With a major NSF award to support our TeraGrid-related operations (NSF Awards $52 Million to PSC for TeraGrid Operations), we look forward to the challenges of moving this ambitious program forward.

Among many scientific accomplishments PSC has enabled, the advance in storm forecasting (Rolling Thunder Preview), high-point of a decade-long collaboration with the Center for Analysis and Prediction of Storms, is at the forefront of work that demonstrates how high-performance computing contributes directly to improving our lives.

Understanding neural processes is an area where PSC marshals special excellence. One feature of PSC’s biomedical program ( The National Resource for Biomedical Supercomputing), which engages HPC expertise with the life sciences, the MCell simulation program, co-authored by PSC scientist Joel Stiles, and supported with our outstanding biomedical visualization capabilities, has proven ability to realistically model extremely complex neurotransmitter events at synapses. Stiles’ recent work on exocytosis (Fire in the Brain) is forcing a rethinking of some long-held views.

Photonic crystals represent a major success story for computational science, as exemplified by two projects reported here (Bringing Light to Heel). Shanhui Fan at Stanford found a new way to stop light in its tracks. The Joannopoulos team at MIT developed a new fiber that makes laser surgeries possible that used to be impossible. This progress depends on the ability to simulate how light behaves, and shows us, if we still need to be shown, that running models inside computers can tell us important things we have no other way to learn.

Ken Jordan’s work (Water, Water Everywhere) with water clusters has led to several papers in Science in the past year, including recognition as one of the 10 most important science breakthroughs of 2004. P.K. Yeung’s work (Taming the Whirlwind) is the largest-scale turbulence simulation yet. Molecular dynamics studies by Lynch and Reggio (All in Your Brain) may help in the development of new therapeutic drugs.

PSC’s development work in many areas, such as PDIO (Big Ben Delivers in Real Time via TeraGrid), facilitates the scientific progress reported here. This is exemplified also by our work with automated checkpoint-restart , which will become more critical as we progress to systems with many more processors. We look forward to collaborating with Keshav Pingali and his team at Cornell to realize the advantages of CPR with Big Ben.

For all these achievements, credit is due to the creative, hard-working staff of PSC. We are most grateful for the support we receive from the National Science Foundation, the U.S. Department of Energy, the National Center for Research Resources of the National Institutes of Health, the Commonwealth of Pennsylvania and many others.