NSF & Congress
Testimony

Mr. Chairman and Members of the Subcommittee, thank you for this opportunity to testify about NSF's new program, Partnerships for Advanced Computational Infrastructure (PACI).

I am Paul Young, Special Advisor to NSF's Directorate for Computer and Information Science and Engineering for the PACI program, and formerly Assistant Director of this NSF Directorate.

The new PACI program, which replaces the successful older NSF Supercomputer Centers program, is an important element in the Foundation's future infrastructure for the support of academic science and engineering, research and education.

A week and a half ago, on March 28th, the National Science Board approved negotiations prior to awards to the National Computational Science Alliance (NCSA), led by the University of Illinois at Urbana/Champaign, and the National Partnership for Advanced Computational Infrastructure (NPACI), led by the University of California, San Diego. This approval represents the formal beginning of the new PACI Program.

This Board action is an important milestone in a process that began two and one half years ago, when in October 1994, the Board extended the current Supercomputer Centers program through FY97. In response to this Board action, the Director of the Foundation appointed a special external Task Force on the Future of the NSF Supercomputer Centers Program, which you just heard about from Dr. Lane.

The Task Force felt, that two major technological factors called for a change in the structure of the existing Centers Program. These were the increasing dominance of scalable parallel computers, with their promise of highly cost effective computing power, and the expected growth and ubiquity of high speed networks.

Networking has become so pervasive in academia that entirely new ways for scientists and engineers to work together are developing. Soon to be available networks, represented by the Next Generation Internet and Internet II, will provide the network bandwidth necessary to begin to eliminate the concept of distance in using computers and related resources, such as massive data archives.

Both parallel computers and high speed networks hold tremendous promise, but both present major intellectual challenges if their full potential is to be realized. The Task Force felt that those researchers who were building and providing the tools and methods for effective and efficient use of these systems, the computer science community, should contribute greatly to the solution of how to better use these new technologies in the solution of large science and engineering problems. They also felt that the scientists and engineers who posed these problems should work closely with the computer scientists.

These breakthrough technologies and intellectual challenges led the Task Force to recommend a new program based on extensive partnering; the partnerships to be selected, as always at NSF, through a rigorous open competition for the best ideas and minds. As a result of the report, after careful analysis and extensive discussion within the Foundation, the staff recommended to the Science Board a competitive solicitation for a new program, Partnerships for Advanced Computational Infrastructure, to be put in place by the beginning of FY98. Fittingly, this solicitation made its debut, on NSF's web site, on December 15, 1995. More than 4000 visits were made to that site in the next 5 months from over 1500 different net addresses.

The proposers followed a torturous review path, even after they had identified their members and drafted their proposals. First came preproposals in April of 1996. These were reviewed by a panel to recommend non-binding mid-course corrections by providing feedback to prospective proposers. Of the ten groups that submitted preproposals, only six took the next step, producing final proposals by Labor Day, after a full summer of collaboration, team building, and writing.

These six proposals were reviewed by another broadly based, expert panel in September of last year; four of the proposals made the cut to be site visited and thus to more fully present their ideas to their peer reviewers.

Finally, another expert panel met just before Christmas, 1996, to go over all of the previous review materials, including the site visit reports. The panel's decision was unanimous, two of the organizations had really "gotten it right", meeting the requirements in the program solicitation with a scope of activity and vision that was compelling. These two new partnerships could well define a new paradigm for how a community of scientists and engineers assembles itself to provide widely accessible infrastructure. In addition, the two successful proposals were highly complementary, forming together, a balanced national program involving some of the best minds and the finest institutions in the country.

After three more months of internal deliberations and analysis, the NSF staff presented a strong case to the Science Board. Despite a preliminary challenge to the process, which the staff effectively rebutted, the Science Board was convinced that, given the vision and scope of the recommended proposals, and given the thoroughness of the review process, the Foundation should negotiate awards to NCSA and NPACI. The Board also approved funds to phase out NSF's support for the current NSF Supercomputer Centers at Pittsburgh and Cornell, convinced that after a transition period, the new program would fully pick up the load and that the new directions were the best way to insure that computation would continue to flourish in the coming environment.

What do these new partnerships bring?

First, and most important, continued access to the most up to date large computers the U. S. can produce, not just at the two main sites, but across the partnerships, with an expected major presence of all U. S. high performance computer manufacturers.

Second, enthusiastic and widespread involvement of the best and brightest in the computer science community, working on software tools for, and problems important to, effective use of these computational engines and developing high-speed networks.

Third, a national education training and outreach plan which can change the way we teach our young children and the way we train a new generation of scientists and engineers, all of whom will increasingly treat computation as an everyday tool.

And finally, and very importantly, the close involvement of the people whose research represents the cutting edge of science. For example:

• Biologists understanding and designing at the molecular level.
  
• Cosmologists testing theories of the origin of the universe.
  
• Engineers designing materials with incredible properties of strength and corrosion resistance.
  
• Atmospheric scientists predicting damaging storms six hours in advance with sufficient accuracy to warn local residents.

In short, computing answers that can change the daily lives of all Americans for the better.

All of this interconnected by a network that will be the envy of the world.

Computation, communications, and information play an increasingly important role in the world we live in. Our nation's world leadership in the 21st century requires leadership, both in the development of these technologies, and in the fundamental research that enables them. The overarching goal of the new PACI program is to provide the advanced computational infrastructure necessary to maintain our country's world leadership in computational science and engineering. The new PACI program is thus a key part of NSF's overall strategy for maintaining world leadership in all of computation, communications, and information. Mr. Chairman, I believe that when this new program is fully operational, the United States will be even better positioned to continue its current world leadership in these critical areas.

See also: Hearing Summary.