Computing Research Policy Blog

Science Committee Views on HPC

The clearest guide to what the House Science Committee intends to accomplish in moving HR 4126 to the House floor for passage is probably found in the report accompanying the bill. I've reprinted the pertinent passages here, as the report doesn't seem to be available yet on Congress' Thomas system.

Interagency Planning and Coordination

The High-Performance Computing Act of 1991 codified an interagency planning process that remains in place today. However, the chief product of this process in recent years has been an annual retrospective review of activities undertaken by agencies, rather than a prospective planning document. The Committee expects all of the participating agencies to engage in a forward-looking planning and coordination process led by OSTP to coordinate high-performance computing activities across the federal government. The agencies, led by OSTP, should submit a coordinated budget for federal high-performance computing activities to the Office of Management and Budget. Furthermore, the agencies, led by OSTP, should develop and periodically refine a research, development, and deployment roadmap for high-performance computing systems. In addition, in formulating plans for the Program, the Committee expects the participating agencies to take into consideration the findings and recommendations of the President’s Information Technology Advisory Committee, which is required to conduct recurring reviews of the planning, implementation, and contents of the Program.

Assuring U.S. Researchers Sustained Access to High-Performance Computing Infrastructure

The Committee believes that the High-Performance Computing Research and Development Program, in general, and NSF and DOE’s Office of Science, in particular, must provide U.S. researchers with sustained access to high-performance computers that are among the most advanced in the world in terms of performance in solving scientific and engineering problems. This is necessary in order for the U.S. to maintain its position as a world leader in scientific and engineering fields and in technology innovation. By “among the most advanced in the world,” the Committee means general purpose scientific computing systems that would rank among the top few systems in existence in performance (1) on widely accepted standardized tests, such as the LINPACK Benchmark used to generate the Top 500 list; and (2) on actual production codes for solving the most demanding problems in science and engineering disciplines. The Committee intends that such computing systems be equivalent to “Leadership Systems” as described in the May 10, 2004 report of the Office of Science and Technology Policy, Federal Plan for High-End Computing.

The Committee is supportive of recent initiatives to make DOE’s Office of Science high-performance computing resources more broadly available to researchers not otherwise supported by DOE and to allocate those resources on a competitive, merit-reviewed basis. The Committee encourages DOE to increase the quantity of supercomputing resources allocated to U.S. researchers in this fashion and to provide information to the research community on the long-term availability of these resources.

The Committee is supportive of continued NSF funding of software, algorithms, networking and data storage techniques, and education and outreach activities associated with high-performance computing. However, the Committee emphasizes that significant attention and funding must also be devoted to procurement of high-performance computing hardware for high-performance computing user facilities, including the NSF supercomputer centers.

Overall, the Committee believes that for the federal government to effectively meet the scientific community’s high-performance computing needs, NSF and DOE’s Office of Science each must support Leadership Systems which should be available for use by researchers from academia, industry, and government laboratories. By use of the phrase “sustained access” the Committee expects NSF and DOE to develop and maintain plans and budgets to assure ongoing improvements in the capability of high-performance computing user facilities, such as the NSF supercomputer centers and DOE’s Office of Science high-end (high-performance) computing user facilities, so that the computing infrastructure made available through these facilities remains among the most advanced in the world.

But the most advanced high-performance computing hardware, on its own, will not be enough to enable researchers to conduct the most advanced science. The Committee believes that the development of software, applications, networking, and data storage and management techniques, including support for the applied mathematics required to develop advanced software and algorithms, will be essential to enable researchers to make effective use of the high-performance computing resources made available under this Act.
National Information Technology Research and Development Program (NITRD)

The NITRD program includes six program component areas: High End Computing, Large Scale Networking, Software Design and Productivity, Human Computer Interaction and Information Management, High Confidence Software and Systems, and Social, Economic, and Workforce Implications of Information Technology. While the focus of this Act is on high-performance computing, the Committee recognizes that all program component areas are essential parts of the federal information technology research and development effort and expects the planning and coordination process for the NITRD program to result in an appropriate balance of resources among the program component areas. The committee expects the annual report for the program to provide the rationale for the allocation of funding among the program component areas. The Committee expects that the allocations for the high end computing program component area will be sufficient to carry out this Act.