Click for next page ( 29


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 28
The Role of Government in Supercomputing The federal government has been involved in the development and advancement of supercomputing since the advent of computers themselves While the precise mechanism and level of support have varied over time, there has been a long-standing federal commitment to encourage the technical progress and diffusion of high-performance computing systems (some of this history is summarized in Chapter 2~. Economic and policy analysis emphasizes several broad justifications for government involvement in technology development; the key justifications that apply to supercomputing are outlined below. GOVERNMENT AS A LEADING CUSTOMER Much technological innovation is (at least initially) directed toward applications dominated by government involvement and purchasing. Most notably, defense needs have often been the specific setting in which new technologies including supercomputing are first developed and applied. Even when commercial firms are purchasing or exploiting supercomputer technology, as in the pharmaceutical and biotechnology industries, governments are often the largest single customer for the resulting innovations (e.g., through the Medicare and Medicaid programs in the United States or national health insurance programs in other countries). The federal government remains the single largest purchaser of supercomputers in the world, for mission-oriented tasks ranging from national security to health to climate modeling. Some government missions may require specifications that are peculiar to the individual applications. In such cases, it may be important for the relevant federal agencies and research labs to be closely involved in the associated R&D (including prototyping), even when the research and development are carried out in the private sector. In the United States, Japan, and elsewhere, the majority of supercomputers have been purchased directly or indirectly using government funds, and the committee has no evidence that this pattern is likely to change in the foreseeable future. As the social custodian of well-defined government missions and the largest and most aggressive customer for new technology related to these missions, the government has an incentive to ensure appropriate and effective funding for innovative investments in IDA estimates that the high-end HPC market in the United States has been around $1 billion (~$200 million) per year since 1994. The U.S. government (including DOD, the national laboratories, and classified programs) spends roughly $700 or $800 million per year (and this spending has been relatively flat for the last 10 years). SOURCE: Debra Goldfarb, IDC HPC industry analyst, presentation to the committee on May 23, 2003. 28

OCR for page 28
THE ROLE OF GOVERNMENTINSUPERCOMPUTING supercomputing technology so as to guarantee that the technology progresses at a rate and in a direction that serve government missions. NATIONAL SECURITY IMPLICATIONS U.S. government users require assured, secure, and reliable access to supercomputing capabilities to ensure that critical national security requirements are met. Supercomputers are used to develop intelligence gathering equipment (e.g., better antennas), to find important information through massive data mining, and for cryptography and cryptanalysts. They are used to design better weapons, airplanes, and tanks as well as for battlef~eld-related calculations that must be carried out very quickly. For example, the timely calculation of areas of enemy territory where enemy radars are not able to spot our airplanes (as was done during the first Gulf war) can be crucial. Design and refurbishment of nuclear weapons depends critically on supercomputing calculations, as does the design of next-generation armament for the Army's Future Combat System. 29 It is likely that supercomputing will be increasingly important to homeland security. Examples include micrometeorology analysis to combat biological terrorism and computer forensic analysis of terrorist bombings. The federal government must be able to guarantee that such systems do what they are intended to do with no harmful side effects from, for instance, malicious insertion of incorrect calculations. It must guarantee that supercomputers are available to U.S. security agencies with no hindrance and must be able to restrict access to some supercomputing technologies abroad. All in all, the government has an Incentive to ensure a strong supercomputing technology base in the United States. MARKET FORCES Economists are generally reluctant to see government intervene in highly competitive markets, where the costs of disruption to well-functioning and efficient resource allocation mechanisms are likely to be high. However, in many circumstances, market-based incentives for scientific discovery and innovation are likely to be insufficient. Because innovators often are unable to capture the full value of their inventions, market forces alone typically will bring less innovation than is worthwhile from society's perspective. Typically, underinvestment is greatest for basic research, fundamental scientific discoveries, or technologies that serve as stepping-stones for follow-on research by others. A number of computing innovations first implemented in supercomputers (for example, instruction lookahead, multiple arithmetic units, multiple instruction buffers and data operators, pipelining, programmable I/O processors)2 played an important role in shaping the architecture and performance of mainstream computers today (from workstations to personal computers). Initiatives funded in the context of supercomputers have influenced the ability to commercialize innovations, from workstation architecture to the latest Intel CPU. Machines equivalent in power to the supercomputer of 15 years ago that cost millions of dollars can now be purchased online for less than $1,000. The dramatic decrease in price for the same performance is due to continuing advances in semiconductor technology. However 2For historical compilations of these and other major innovations in computer architecture, see Harold S. Stone et al., 1980, "Hardware Systems," in What Can Be Automated? The Computer Science and Engineering Research Study, Bruce W. Arden, ea., Cambridge, Mass.: MIT Press, pp. 3 1 9 and 320; C. Gordon Bell and Allen Newell, 1971, Computer Structures: Readings and Examples, New York, N.Y.: McGraw-Hill, fig. 2c, p. 45, p. 71; C. Gordon Bell and Allen Newell, 1982, Computer Structures: Principles and Examples, New York, N.Y.: McGraw- Hill, p.393. These and other sources have been drawn together and annotated with additional information in K. Flamm, 1988, Creating the Computer: Government, Industry, and High Technology, Washington, D.C.: Brookings Institution Press. pp. 260-269.

OCR for page 28
30 THE FUTURE OF SUPERCOMPUTING: ANINTERIMREPORT the architecture of the chip and the software that drives it are based on the supercomputers of 15 years ago. Supercomputing technology created with government funds has found its way into commercial computers and, in turn, has been used by private firms in basic research and to improve the design of products, including airplanes, automobiles, and pharmaceuticals. Supercomputing is essential for fundamental and applied research in materials science, earth science, life sciences, and so on. It is interesting to note that the Earth Simulator system that is described earlier is dedicated exclusively to the geosciences. The availability of such a powerful too! is likely to accelerate research in those disciplines by leading, for example, to a better understanding (and, hence, better prediction) of climate change and earthquakes. The value of such accelerated progress appears to be immense. It would seem, however, that only a small share of all the benefits is being captured by those involved in the initial stages of their development. In summary, keeping the U.S. supercomputing industry at the technological cutting edge is vital for the security interests of the United States. In light of this imperative, there are several economic arguments that might justify continued government funding for supercomputing R&D rather than reliance on the marketplace alone.