NSF & Congress
Dr. Neal Lane
National Science Foundation
Before the House Science Committee
Subcommittee on Basic Research
April 9, 1997
Chairman Schiff and members of the committee, I am pleased to have the opportunity to provide the committee with some additional background on how the NSF Partnership for Advanced Computational Infrastructure or PACI program came about.
The predecessor to PACI, the NSF Supercomputer Centers Program, was initiated in the mid 1980s with the formation of five supercomputer centers. These centers, preeminent for that time period, provided vector supercomputing services and training for the research community. After five years of initial NSF support, four of these centers were approved for another five years. During this second phase, there was a vigorous effort to pursue parallel computing and to expand the outreach efforts of the centers. These centers formed themselves into a MetaCenter, sharing resources among themselves and building linkages with industry.
So, Mr. Chairman, it was after ten years of a successful program that, in October of 1994, the National Science Board asked whether NSF should continue support for the current supercomputing centers program or phase out the existing program to make room for a new one. To answer this question, I appointed a committee, chaired by Ed Hayes, Vice President for Research at Ohio State University. The Hayes Task Force was comprised of high performance computing experts from academia, industry and government with exceptional talent and breadth of experience.
From January to September 1995, the Hayes Task Force evaluated the success of NSF's existing program and examined the future needs of the research community for high performance computing. As part of its background research, the Task Force utilized the 1993 Report of the NSF Blue Ribbon Panel on High Performance Computing, chaired by Lewis Branscomb, and the 1995 National Research Council's report entitled, "Evolving the High Performance Computing and Communications Initiative to Support the Nation's Information Infrastructure."
The Task Force also gathered substantial input from industry, academia, government, and government laboratories. Over 500 people replied to its survey on the effectiveness of the current program and the future of U.S. supercomputing. The extensive participation by the community indicates clearly the importance and impact of the NSF Centers Program. Thus, when the Task Force concluded that leading edge computing had made significant contributions to the advance of U.S. science and engineering leadership and strongly recommended that a new program be established, the research community was not surprised.
Dr. Hayes presented the Task Force findings to the National Science Board in October 1995, and the Board approved the PACI program in December 1995. The Hayes Task Force Report's vision of the future of U.S. supercomputing proposed that leading edge sites that maintain the highest end computational systems build partnerships with selected research centers. In order to realize this vision, the Task Force proposed that NSF announce a new competition for a restructured High Performance Computing Centers program that would permit funding of selected sites for a period of five years.
It recommended two major, and interrelated, changes to the old NSF Supercomputer Centers Program. First, a new program should support national "leading-edge sites" with a balanced set of high-end hardware and software capabilities, coupled with appropriate staff needed for continued rapid advancement in computational science and engineering. Second, each leading-edge site should be partnered with experimental facilities at universities, NSF research centers, and/or national and regional high performance computing centers. Finally, it urged that the "new " program (PACI) should support the needs of the national computational science community through leading edge sites and their partners, rather than through independent basic research.
Mr. Chairman, the report emphasized that the computational capability of the leading edge centers should be one or two orders of magnitude beyond what is available at leading research universities. To achieve such economies of operation and to maintain the very high end capability, it was clear that a reduction in the number of sites would likely be necessary. The Task Force stated clearly, and I quote,
These recommendations were designed to set the Centers program on a new course that builds on its past successes, yet shifts the focus to the present realities of high-performance computing and communications, and provides flexibility to adapt to changing circumstances. It is our expectation that at current NSF budget levels and absent new outside resources, there will be a reduction in the number of leading-edge sites to effect the benefits of the Task Force recommendations.
My senior staff and I reviewed the Task Force recommendations, carefully considering the potential implications of the suggested improvements to the program. We believe we are taking the right steps to ensure U.S. leadership in high performance computing. In the words of Dr. Diana Natalicio, Vice Chair of the National Science Board, "We sometimes need to make the difficult choice to stop doing something good to enable something great."
Mr. Chairman, as you and the other members of the Subcommittee may recall from last year's hearing on the NSF Supercomputer Centers Program, there has been substantial support among the research community for this new program. At last year's hearing, Dr. Hayes expressed great pleasure at the manner in which the National Science Foundation carried out the recommendations of the Task Force.
Before I turn to Dr. Paul Young, who will provide the details of how NSF carried out the Task Force proposal, I want to emphasize to the Subcommittee that the NSF Partnership for Advanced Computational Infrastructure, or PACI, is not simply a continuation of our expiring Supercomputer Centers Program. Rather, in accordance with the recommendations of the Hayes Task Force and the guidance provided by the National Science Board, PACI goes well beyond the current paradigm of supercomputing centers. It is carefully designed to build the infrastructure needed for both the education and training of future generations of world leaders in science and technology and for the more effective use of new parallel supercomputers by researchers today.
Mr. Chairman, thank you for this opportunity to discuss the extensive NSF review process that went into establishing our new PACI program. I would be happy to respond to any questions that you or other members of the committee might have.
See also: Hearing Summary.