DOE Announces 51 Awards in New Advanced Computing ProgramSeptember 10, 2001
On August 14, the U.S. Department of Energy announced the first awards under its Scientific Discovery through Advanced Computing (SciDAC) program. Fifty-one projects will receive a total of $57 million this fiscal year to advance fundamental research in areas related to DOE's missions, including climate modeling, high-energy physics, and high-performance computing.
SciDAC researchers will create a new generation of scientific simulation codes that will take advantage of the capabilities of terascale computers to address ever larger, more complex problems. The program includes research on improved mathematical and computing systems software that will allow these codes to run efficiently on modern parallel computers. The program will also develop "collaboratory" software that will enable geographically separated scientists to work together as teams; it will also help scientists to control scientific instruments remotely and to share data more readily.
"SciDAC is bringing together integrated teams that consist of the top computational scientists and the best applied mathematicians and computer scientists to work together on scientific simulations in the fields of computational climate dynamics, high-energy and nuclear physics, fusion energy, and computational chemistry," says Charles Romine, a program manager in DOE's Mathematical, Information and Computational Sciences (MICS) Division. "It is generating a lot of excitement because the potential impact of SciDAC on the field of high-performance computational science is enormous!"
The SciDAC activities include 23 large projects, each to be funded at $500,000 to $4 million per year for three to five years, and 27 smaller projects (up to $500,000 per year for three years). The projects involve collaborations among 13 DOE laboratories and more than 50 colleges, universities, and companies. Three components of the SciDAC initiative reside in the MICS Division (which is part of the DOE's Office of Advanced Scientific Computing Research):
- Applied Mathematics Integrated Software Infrastructure Centers
- Computer Science Integrated Software Infrastructure Centers
- National Collaboratory Software Environment Development Centers
As always, a new program has new terminology and acronyms, and SciDAC is no exception, with Integrated Software Infrastructure Centers (ISICs), further divided into Applied Mathematics ISICs and Computer Science ISICs. To give readers an idea of the scope of the projects, here's a glimpse of one ISIC in each of the two categories:
The Terascale Optimal PDE Solvers ISIC, with Old Dominion University as the lead institution and David Keyes as the main contact (and one of several PIs), will focus on the development and implementation of optimal or near-optimal schemes for PDE simulations and closely related tasks, including optimization of PDE-constrained systems, eigenanalysis, and adaptive time integration. The TOPS ISIC will research, develop, and deploy an integrated toolkit of open source, (nearly) optimal complexity solvers for the nonlinear PDEs that arise in many application areas, including fusion, accelerator design, global climate change, and reactive chemistry. These algorithms, primarily multilevel methods, aim to reduce computational bottlenecks by one to three orders of magnitude on terascale computers, making scientific simulation possible on a scale heretofore impossible. Along with usability, robustness, and algorithmic efficiency, an important goal will be to attain the highest possible computational performance in implementations by accommodating to the memory-bandwidth limitations of hierarchical memory architectures.
Participating institutions, in addition to ODU, are Argonne, Lawrence Berkeley, and Lawrence Livermore National Laboratories; the University of California, Berkeley; Carnegie Mellon University; the University of Colorado; New York University; and the University of Tennessee.
Oak Ridge National Laboratory is the lead institution of the Scalable Systems Software ISIC, which will address the lack of software for effective management and utilization of terascale computational resources; ORNL's Al Geist is a PI and the main contact. The virtual center will be a multi-institution, multidisciplinary group composed of experts working together to develop an integrated suite of machine-independent, scalable systems software components needed for the SciDAC program. The goal is to provide open source solutions that work on a range of systems, from small- to large-scale.
Joining ORNL are Argonne, Los Alamos, Lawrence Berkeley, Pacific Northwest, and Sandia National Laboratories; and the University of Illinois, Urbana-Champaign. Along with the ISICs are several "National Collaboratory Software Environment Development Centers": Two "middleware" projects are focusing on individual technology elements that will allow universal, ubiquitous, easy access to remote resources and that will facilitate the work of distributed research teams, and four "collaboratories" are investigating enabling technologies for distributed scientific applications.
Additional information about SciDAC can be found at http://www.sc.doe.gov/. Accounts of progress in specific projects will appear in SIAM News in coming months.