Page 97

C
Review of the High Performance Computing and Communications Initiative Budget

The committee attempted a nontraditional look at how High Performance Computing and Communications Initiative (HPCCI) funds are being invested.' Traditional HPCCI budget reports show budget breakdowns by agency and by program component (High-Performance Computing Systems, National Research and Education Network, Advanced Software Technology and Algorithms, Information Infrastructure Technology and Applications, and Basic Research and Human Resources). The committee found it informative to examine the funding from a functional perspective to understand what sort of technical work is being performed and in what quantity. The committee separated the 88 HPCCI program elements into 11 disciplines defined as indicated below:

· Computer technology (CPT)—applied research directed at advancing the state of computer architecture and hardware technology;

· Software technology (SWT)—applied research directed at advancing the state of computer software technology;

· Communications technology (CMT)—applied research directed at advancing the state of communications technology;

· Computing infrastructure (CPI)—acquisition and operation of supercomputer facilities;

· Communications infrastructure (CMI)—acquisition and operation of high-performance computer communications networks and services;

· Applications and computational science (APP)—creation of software and computational techniques directed at solving specific scientific problems and applications;

· Common applications support (CAS)—creation of software and computational techniques to support a range of applications across multiple disciplines;

· Artificial intelligence and human-machine interaction (AI)—applied research directed at solving artificial intelligence and human interface problems.

Page 98

· Basic hardware technology (BHW)-basic electronics research supporting electronic components that might be applied to a wide variety of systems, including computers and communications systems;

· Education (EDU)—training and education; and

· Administration (ADM)—National Coordination Office.

The committee classified 86 of the 88 program elements as coming under I of the 11 disciplines listed above, based on each program element's FY 1995 milestones (Table C.1). If a program element appeared to fit into more than one discipline, the committee categorized it by examining the element's milestones to determine where the majority of the program activity was concentrated. Two of the larger Advanced Research Projects Agency (ARPA) program activities (Intelligent Systems and Software, and Information Sciences) were split between two disciplines.

Table C.2 shows the FY 1993 actual budget, the FY 1995 request,2and the percentage change in the HPCCI budget for each of these 11 disciplines.

BUDGET REVIEW

It is interesting to examine the HPCCI budget to see which areas are being emphasized and to compare these with the HPCCI's goals and objectives. As indicated also in Chapter 2, the current program goals are as follows:

· Extend U.S. leadership in high-performance computing and networking technologies;

· Disseminate the technologies to accelerate innovation and serve the economy, national security, education, and the environment; and

· Spur gains in U.S. productivity and industrial competitiveness.

The computer technology, software technology, and communications technology disciplines address the goals of extending technical leadership in computing and communications and providing key enabling technologies for the information infrastructure. The budget for these three disciplines accounted for 32.9 percent of the FY 1993 actual budget and 30.5 percent of the FY 1995 requested budget.

The largest part of the HPCCI budget is invested in applications and supercomputer computing infrastructure to support applications—49.8 percent of the FY 1993 actual budget and 50.1 percent of the FY 1995 requested budget. Applications and computational science, common applications support, artificial intelligence and human-machine interaction, and computing infrastructure programs are included. The rest of the budget requested for FY 1995 is divided among basic hardware technology (5.1 percent), communications infrastructure (7.9 percent), education (5.6 percent), and a very small amount for program administration.

Page 99

TABLE C.1 Mapping of Agencies, HPCCI Budget Allocations, Program Elements, and Discipline Categories

Page 100

TABLE C. 1—continued

Page 101

TABLE C.1—continued

Page 102

TABLE C. 1—continued

^AARPA, Advanced Research Projects Agency; NSF, National Science Foundation; DOE, Department of Energy; NASA, National Aeronautics and Space Administration; NIH, National Institutes of Health; NSA, National Security Agency; NIST, National Institute of Standards and Technology; NOAA, National Oceanographic and Atmospheric Administration; EPA, Environmental Protection Agency.

^BAl, artificial intelligence and human-machine interaction; SWT, software technology; CPT, computer technology; BHW, basic hardware technology; CMT, communications technology; EDU, education and training; APP, applications and computational science; CAS, common applications support; CPI, computing infrastructure; CMI, communications infrastructure; ADM, National Coordination Office.

^CNC/GC, National Challenge/Grand Challenge; DCRT, Division of Computer Research and Technology (NIH); IAIMS, Integrated Academic Information Management System; NLM, National Library of Medicine; NCRR, National Center for Research Resources (NIH); NCI, National Cancer Institute (NIH); PDQ, Physician Data Query (NIH).

SOURCE: Data on agency budgets and program activities were extracted from the FY 1995 Implementation Plan prepared by the National Coordination Office (1994).

Page 103

TABLE C.2 Actual FY 1 993 and Requested FY 1995 HPCCI Budget (millions of dollars) Categorized by Discipline

Alternatively, the 11 discipline categories can be used to examine the balance between support for discipline-specific scientific research that uses high-performance computing and communications technologies and support for computer science research on new high-performance computing and communications technologies. Analysis of the FY 1995 HPCCI budget request shows that $352 million (30 percent) would be invested in basic research in computer, software, and communications technologies; $205 million (18 percent) in applied computer science research, artificial intelligence, and human-machine interaction; $176 million (15 percent) in direct support of applications and computational science; and $297 million (26 percent) in computing and communications infrastructure.

Commentary: Many Possibilities for Misinterpretation

The HPCCI has enjoyed a certain amount of political support and is growing even in a time of very tight federal budgets. The committee believes that this has created a ''bandwagon" effect: the initiative has had its scope extended by the inclusion of some work not directly related to the HPCCI's goals, however valuable it may be, or work with broad relevance. The result has been a less than focused program.

For example, all high-performance computing and communications systems are built from electronics and depend directly on advancements in basic electronic technology. The ARPA Microsystems program activity, which constitutes the large majority of the basic hardware discipline, supports research in basic electronics technologies. This research will eventually benefit

Page 104

the high-performance computing and communications technology base and help advance the nation's information infrastructure, but it could also be used in a wide variety of other contexts.

Another problem is the possible creation of false expectations about the extent to which the HPCCI could create the technology necessary for advancing the nation's information infrastructure. A large amount of work within the two applications disciplines is directed primarily toward the use of high-performance computing in solving certain scientific and agency mission problems. Only a part of this work, such as the creation of digital libraries, would apply directly to the goal of enhancing the nation's information infrastructure. Some of this work is directed at challenging computational science problems, which have excellent scientific impact but whose results are more easily justified as scientific results, rather than HPCCI results. Also, the HPCCI invests much more heavily in computing than in communications. Less than 16 percent of the FY 1995 request is for communications technology and infrastructure.

About one-third of the program is directed toward creating new technology directly applicable to advancing the information infrastructure. The growth in funding in these areas is offset by an unrelated decrease in research investment by industry, spurred in part by competitive changes in the computer and communications industries. As a result, the nation's total amount of research in high-performance computing and communications technologies is considerably less than it appears, and in fact may be insufficient to maintain the strategic U.S. lead in these technologies or to support the rapid deployment of an enhanced information infrastructure.

NOTES

1. CPSMA (1994), p. 7; this report points out that labor-intensive, detailed disaggregation of published data may be the only way to understand how research program budgets are spent.

2. Amounts shown for FY 1995 are Executive Budget requests. At press time, agency appropriations had been made, but the involved agencies had not disaggregated the appropriations and reported the HPCCI portions to the National Coordination Office. A 2-year time period, FY 1993 to FY 1995, was used to help dampen any single-year jumps in level.