High Performance Computing and Communications: New Program Direction
Would Benefit from a More Focused Effort (Chapter Report, 11/04/94,
GAO/AIMD-95-6).

The federal High Performance Computing and Communication program seeks
to accelerate research and development of high performance computers and
networks and promote their use in both the federal government and the
private sector.  If successful, the program could significantly extend
U.S. technological leadership and enhance national competitiveness.
This report examines (1) the effectiveness of the program's management
structure in setting goals and measuring progress and (2) how
extensively private industry has been involved in the program's planning
and execution.  GAO also identifies opportunities to improve the program
and makes recommendations to focus the program on its new role in
developing new technology in support of the national information
infrastructure.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  AIMD-95-6
     TITLE:  High Performance Computing and Communications: New Program 
             Direction Would Benefit from a More Focused Effort
      DATE:  11/04/94
   SUBJECT:  Information technology
             Telecommunication industry
             Computer equipment industry
             Research and development
             Command/control/communications/computer systems
             Research program management
             Computer modeling
             Systems compatibility
             Information resources management
             Systems conversions
IDENTIFIER:  High Performance Computing and Communications Program
             ARPA High Performance Computing Program
             Computer Systems Policy Project
             
*******************************************************************************
*   This file contains an ASCII representation of the text of a GAO           *
*   report.  Delineations within the text indicating chapter titles,          *
*   headings, and bullets are preserved.  Major divisions and subdivisions    *
*   of the text, such as Chapters, Sections, and Appendixes, are              *
*   identified by double and single lines.  The numbers on the right end      *
*   of these lines indicate the position of each of the subsections in the    *
*   document outline.  These numbers do NOT correspond with the page          *
*   numbers of the printed product.                                           *
*                                                                             *
*   No attempt has been made to display graphic images, although figure       *
*   captions are reproduced. Tables are included, but may not resemble        *
*   those in the printed version.                                             *
*                                                                             *
*   A printed copy of this report may be obtained from the GAO Document       *
*   Distribution Facility by calling (202) 512-6000, by faxing your           *
*   request to (301) 258-4066, or by writing to P.O. Box 6015,                *
*   Gaithersburg, MD 20884-6015. We are unable to accept electronic orders    *
*   for printed documents at this time.                                       *
*******************************************************************************


Cover
========================================================================== COVER


Report to the Chairman, Committee on Armed Services, House of Representatives

November 1994

HIGH PERFORMANCE COMPUTING AND
COMMUNICATIONS - NEW PROGRAM DIRECTION
WOULD BENEFIT FROM A MORE FOCUSED
EFFORT

GAO/AIMD-95-6

High Performance Computing


Abbreviations
========================================================================= ABBREV

  ARPA - Advanced Research Projects Agency
  DOD - Department of Defense
  DOE - Department of Energy
  EPA - Environmental Protection Agency
  GAO - General Accounting Office
  HPCC - High Performance Computing and Communications
  HPCCIT - High Performance Computing, Communications, and Information
     Technology
  IITA - Information Infrastructure Technology and Applications
  MPP - massively parallel processing
  NASA - National Aeronautics and Space Administration
  NCO - National Coordination Office for HPCC
  NIH - National Institutes of Health
  NII - national information infrastructure
  NIST - National Institute of Standards and Technology
  NOAA - National Oceanic and Atmospheric Administration
  NSA - National Security Agency
  NSF - National Science Foundation
  NSTC - National Science and Technology Council
  OSTP - Office of Science and Technology Policy

Letter
========================================================================= LETTER


B-257448

November 4, 1994

The Honorable Ronald V.  Dellums
Chairman, Committee on Armed Services
House of Representatives

Dear Mr.  Chairman: 

This report responds to your request that we evaluate the status of the federal
High Performance Computing and Communications (HPCC) program.  Specifically,
you asked that we assess (1) the effectiveness of the program's management
structure in setting goals and measuring progress and (2) how extensively
private industry has been involved in the planning and execution of the
program.  The report identifies opportunities to improve the program and makes
recommendations to the Director of the Office of Science and Technology Policy
to focus the program more explicitly on its new role in developing new
technology in support of the national information infrastructure. 

We are sending copies of the report to appropriate congressional committees,
the Director of the Office of Science and Technology Policy, the Secretaries of
the Departments of Defense and Energy, the Administrator of the National
Aeronautics and Space Administration, and the Directors of the National Science
Foundation and the National Coordination Office for High Performance Computing
and Communications.  Copies will also be made available to others upon request. 

This work was performed under the direction of David O.  Nellemann, Director,
Information Resource Management/National Security and International Affairs,
who can be reached at (202) 512-2666.  Other major contributors are listed in
appendix II. 

Sincerely yours,

Gene L.  Dodaro
Assistant Comptroller General


EXECUTIVE SUMMARY
====================================================================== Chapter 0


   PURPOSE
-------------------------------------------------------------------- Chapter 0:1

The federal High Performance Computing and Communications (HPCC) program aims
to accelerate research and development of high performance computers and
networks and promote the use of those resources in both the federal government
and the private sector.  A successful HPCC program could significantly extend
U.S.  technological leadership and enhance national competitiveness. 

Given the potential impact of the HPCC program on both government and industry,
the House Armed Services Committee asked GAO to examine (1) the effectiveness
of the program's management structure in setting goals and measuring progress
and (2) how extensively private industry has been involved in the planning and
execution of the program. 


   BACKGROUND
-------------------------------------------------------------------- Chapter 0:2

The HPCC program was first included in the President's budget in fiscal year
1992 as a coordinated effort among nine federal agencies to accelerate the
availability and utilization of the next generation of high performance
computers and networks.  The program was also specifically authorized by
Congress in the High Performance Computing Act of 1991 (Public Law 102-194). 
The program, coordinated through the White House Office of Science and
Technology Policy (OSTP), was initially funded at $654 million in fiscal year
1992.  Funding for fiscal year 1995 is projected to grow to more than $1
billion. 

The HPCC program grew out of successful ongoing computer and communications
research programs at participating agencies.  HPCC and its predecessor agency
programs were instrumental in establishing more than a dozen high performance
computing research centers throughout the U.S.  Efforts to provide nationwide
access to these centers through interconnected high-speed data networks have
led to dramatic increases in the use of these networks.  The computing research
centers and networks have, in turn, allowed scientists to make significant
advances in addressing the highly complex, scientific problems that are
collectively referred to as "grand challenges." Grand challenges include such
problems as understanding global climate change, analyzing nuclear reactions,
and mapping the human genetic structure. 

In 1993, the administration expanded the scope of the HPCC program to include a
broader range of applications that will have a more direct, near-term impact on
the national information infrastructure (NII), also known as the "information
superhighway." Although it will be built and operated by the private sector,
the NII will involve a wide range of government networking and applications
projects in addition to HPCC, which will serve as the NII's research and
development laboratory.  Proponents envision the NII as a large, interconnected
resource of computers and communications networks that will enhance information
access and delivery and that will be essential to the nation's economic
competitiveness.  Within this context, HPCC research is intended to improve
computerized support for areas that affect all Americans, such as health care,
education, and manufacturing. 


   RESULTS IN BRIEF
-------------------------------------------------------------------- Chapter 0:3

Much valuable research has already been accomplished within the context of the
HPCC effort.  Participating agencies have sponsored highly successful high
performance computing research centers and networks that have allowed
scientists to make significant progress in addressing complex problems in a
variety of scientific and engineering disciplines. 

The administration is now broadening the role of the HPCC program in developing
new technology in support of the NII.  Industry and academic researchers agree
that specific technology areas will need to be targeted for development to
support the NII.  This shift in priorities will require changes in planning to
accommodate these needs.  Given this new context, a more focused management
approach could help better ensure that the program's goals are met.  A detailed
technical agenda will be needed to identify priority areas and commit resources
to them.  Budget information is not prepared consistently from agency to
agency, which has also reduced visibility into how the government is currently
investing in HPCC. 

Industry participation, which has always been crucial to HPCC, is even more
important now that the administration has asserted HPCC's link to the planned
NII.  Industries that could capitalize on HPCC technologies to create new
products and services for the NII could be better represented among HPCC
program participants. 


   PRINCIPAL FINDINGS
-------------------------------------------------------------------- Chapter 0:4


      NEW PROGRAM DIRECTION WOULD
      BENEFIT FROM A MORE FOCUSED
      MANAGEMENT APPROACH
------------------------------------------------------------------ Chapter 0:4.1

The HPCC program was originally established, by design, as a loosely
coordinated, scientifically oriented research effort rather than a rigorously
managed development program.  Now, however, the administration is counting on
HPCC to develop the new technology that will be needed to make the NII
successful.  Given that resources for research and development are limited,
industry and academic researchers agree that specific technology areas will
need to be targeted for development to support the NII. 

The HPCC program will have to develop a detailed technical agenda as a
framework for guiding the government's investment in HPCC research.  This
technical agenda would serve as a master program plan, identifying and
prioritizing specific technical challenges and establishing a framework of
expected costs and results, so that program progress could be measured and
costs controlled.  The HPCC program's fiscal year 1995 Implementation Plan,
which provides summary information about planned activities at each of the
participating agencies, does not yet fulfill this need because it does not
present a prioritized agenda of research areas linked to the needs of the NII
or rationalize its allocation of funds to each of the supported research areas. 

HPCC budgets are currently developed by managers at each participating agency
without any formal guidelines for the program as a whole.  As a result, the
kinds of research activities included in the program and how they are
categorized vary significantly among participating agencies, making it
difficult to determine how the government is apportioning funds among competing
HPCC research areas.  The Implementation Plan is a step in the right direction
in that it sets a standard format for presenting budget information and
presents more detailed information than has previously been publicly available. 
Nevertheless, a more standard method for characterizing HPCC spending across
agencies would afford even greater visibility into the overall federal
investment and facilitate more informed assessments of whether appropriate
emphasis is being placed on areas that need greatest attention.  More open and
consistent reporting of funding could also broaden industry support for the
program by clarifying the program's major interest areas and funding
priorities. 


      GREATER INDUSTRY INVOLVEMENT
      COULD HELP
------------------------------------------------------------------ Chapter 0:4.2

Close collaboration with a broad range of industries is essential to ensuring
that the HPCC program meets its goal of accelerating the development and
widespread use of HPCC technologies.  Although the High Performance Computing
Act mandated appointing an advisory committee, including representatives from
industry, concerns about potential conflict-of-interest have slowed the effort
to establish one.  Establishing partnerships with industry has become even more
important now that the administration has linked HPCC to the planned NII. 
Industries that could capitalize on HPCC technologies to create new products
and services for the NII could be better represented among HPCC program
participants.  For example, there has been little opportunity for involvement
by potential developers of new software applications that would make it easier
for the average person to interact with the NII.  The program's National
Coordination Office (NCO), which has made some progress in involving industry,
could help arrange more opportunities for industry representatives to provide
substantive input to the program. 

Industry representatives interviewed by GAO generally expressed interest in
participating in the HPCC program but saw the program as not being designed to
accommodate them.  While there has been some industry participation, industry
was not invited to participate in developing program plans.  Consequently,
several key industry concerns have not been adequately addressed, including the
need to emphasize applications, software development, and standards-setting
activities. 


   RECOMMENDATIONS
-------------------------------------------------------------------- Chapter 0:5

The Director of OSTP should take certain actions to focus the HPCC program more
explicitly on its new role in support of the NII and to involve industry more
closely in planning and executing the program.  For example, GAO recommends
that OSTP direct the HPCC program managers to develop an explicit HPCC
technical agenda that delineates the program's overall goals, objectives, and
development strategy and that sets priorities and measures for specific
technology areas.  GAO also recommends that OSTP direct the Director of the NCO
to take additional steps to promote industry participation, including involving
industry representatives in the program planning process, and to provide
greater support for standards-setting activities. 


   AGENCY COMMENTS AND GAO'S EVALUATION
-------------------------------------------------------------------- Chapter 0:6

In a September 1994 letter, the Assistant to the President for Science and
Technology (Science Advisor) generally concurred with GAO's findings and
recommendations.  His comments are reprinted, along with GAO's evaluation, in
appendix I.  Specifically, the science advisor agreed that a more focused
management approach is appropriate, given the new direction of the program, and
that a more detailed and prioritized technical agenda is called for along with
improved consistency in preparation of HPCC budgets.  He also strongly
concurred with the recommendation that a private sector advisory committee be
established. 

The science advisor disagreed with what he perceived as GAO's view that the
program be centrally managed and that it have a centrally controlled budget. 
GAO did not recommend centralizing the program's management or budget.  GAO
believes that HPCC program goals can be met within the framework of the
existing program structure.  However, achieving and sustaining the kind of
expanded effort now envisioned for HPCC will require identifying specific
technical goals and priorities and establishing a clear framework for deciding
the type of activities to be funded within the program.  A committee of the
National Research Council issued a July 1994 interim report on the HPCC program
that raised concerns in many of the same areas that GAO addressed, including
the need for effective performance measures, budget consistency, greater
emphasis on software, and greater industry involvement. 


INTRODUCTION
====================================================================== Chapter 1


   BACKGROUND
-------------------------------------------------------------------- Chapter 1:1

The federal High Performance Computing and Communications (HPCC) program began
in fiscal year 1992 as a joint effort among nine federal agencies to
significantly accelerate the availability and utilization of the next
generation of high performance computers and networks.  The overall goals of
the program are to

extend U.S.  technological leadership in high performance computing and
computer communications;

provide wide dissemination and application of the technologies to speed the
pace of innovation and to improve national economic competitiveness, national
security, education, health care, and the global environment; and

provide key parts of the foundation for the national information infrastructure
(NII) and demonstrate selected NII applications. 

Four agencies--the Advanced Research Projects Agency (ARPA), the Department of
Energy (DOE), the National Aeronautics and Space Administration (NASA), and the
National Science Foundation (NSF)--developed the original program plan for HPCC
in 1989, and they remain the program's dominant participants.  In fiscal year
1995 these agencies together will spend more than $900 million, or 81 percent
of the official budget.  Led by the White House Office of Science and
Technology Policy (OSTP), programs at each of these agencies were drawn
together to form the governmentwide HPCC program.  Ten federal agencies
currently participate in the HPCC program.\1

OSTP has designated the National Science and Technology Council (NSTC) to
oversee the HPCC program, through its Committee on Information and
Communications.  The NSTC, a cabinet-level organization created by the
President in November 1993, is intended to serve, in part, as a mechanism for
coordinating research and development strategies across the government and for
monitoring agency research and development spending plans.  Since the NSTC has
only recently been established, it is still too early to gauge its impact on
the HPCC program.  Table 1.1 presents an overview of reported HPCC spending to
date and budgeted amounts for fiscal year 1995 by participating agency. 



                          Table 1.1
           
              HPCC Program Funding by Agency and
                         Fiscal Year

                    (Dollars in millions)

                                                1995
                        1992    1993    1994  Reques
Agency        Base\a  Actual  Actual    Est.       t   Total
------------  ------  ------  ------  ------  ------  ======
ARPA             183     232     275     299     357   1,346
NSF              169     201     225     267     329   1,191
DOE               65      92     101     123     125     506
NASA              54      71      82     113     125     445
NSA               44       -       -      42      40     126
NIH               14      41      47      58      82     242
NOAA               1      10      10      11      25      57
EPA                1       5       8       7      14      35
NIST               2       2       2      18      56      80
============================================================
Total            533     654     750     938   1,153   4,028
------------------------------------------------------------
Source:  National Coordination Office. 

Note:  We did not independently verify the reported funding amounts. 

\a Except for NSA, the HPCC base represents fiscal year 1991 funding for
projects that became the core of the official program in fiscal year 1992. 
When NSA officially joined the program in fiscal year 1994, it identified $44
million in ongoing fiscal year 1993 activities that were HPCC-related.  These
funds can be considered the base funding for NSA's program. 

The 1989 program plan laid out the original framework and parameters for the
government's investment in HPCC, proposing that the program grow in even
increments from a base of approximately $500 million to approximately $1.1
billion in its fifth year.  In budgeting for the actual program, HPCC managers
have adhered closely to these original targets.  Spending is anticipated to
continue at over $1 billion annually until 1998. 

To date, the HPCC program and its predecessor agency programs have been highly
successful.  Participating agencies have been instrumental in establishing more
than a dozen high performance computing research centers throughout the U.S. 
Efforts to provide nationwide access to these centers through interconnected
high-speed data networks have led to dramatic increases in the use of those
networks.  The computing research centers and networks have, in turn, allowed
scientists to make significant advances in addressing the highly complex,
scientific problems that are collectively referred to as "grand challenges."
Grand challenges include such problems as understanding global climate change,
analyzing nuclear reactions, and mapping the human genetic structure. 

In September 1992, OSTP established a National Coordination Office (NCO) to
coordinate the activities of the agencies participating in HPCC and to serve as
liaison to Congress, industry, academia, and the public.  The office's director
serves part-time; this individual is also director of the National Library of
Medicine.  The NCO provides administrative support, disseminates information,
and chairs coordination meetings attended by officials of the participating
agencies.  The NCO does not assess agency HPCC programs or provide guidance to
the agencies on their programs.  It also does not review or have approval
authority regarding agency HPCC budgets. 


--------------------
\1 In addition to the four original participating agencies, the other six
agencies are the National Security Agency (NSA), the National Institutes of
Health (NIH), the National Oceanic and Atmospheric Administration (NOAA), the
Environmental Protection Agency (EPA), the National Institute of Standards and
Technology (NIST), and the Department of Education.  NSA joined the HPCC
program in fiscal year 1994.  The Department of Education has been
participating since fiscal year 1992, but does not have any funding designated
for HPCC activities. 


   AGENCY HPCC PROGRAMS VARY
   CONSIDERABLY
-------------------------------------------------------------------- Chapter 1:2

Since HPCC is structured as a consortium of federal research agencies with
independent programs and budgets, participating agencies can--and do--have
widely varying approaches to research and development.  The programs of the
four major participants reveal the diversity of these agencies' approaches. 
ARPA and NSF--the major participants in terms of expenditures--are quite
different from NASA and DOE.  ARPA and NSF concentrate more heavily on basic
research, although all four agencies fund scientists working on practical
applications of HPCC technologies. 

ARPA has been at the forefront of research into critical technologies, such as
computer time-sharing, computer graphics, computer networks, and artificial
intelligence, for many years.  The agency has had a high performance computing
program since the early 1980s.  ARPA funds some 200 or more HPCC projects, most
of which are relatively small-scale efforts costing between $100,000 and
$500,000.  Having no laboratories or centers of its own, ARPA funds projects
that are run half by academic researchers and half by industry and other
government researchers.  It also funds the placement of HPCC computers and
networks at research sites for use on a variety of research problems. 

NSF, like ARPA, funds a large number of relatively small-scale research
projects in a wide range of scientific disciplines.  NSF also is similar to
ARPA in providing HPCC computing and communications infrastructure for a range
of research uses.  NSF does this by providing base funding for four national
supercomputer centers that, in turn, support research in a range of
disciplines, such as biotechnology, global change studies, and manufacturing
design. 

NASA and DOE, in contrast to ARPA and NSF, are involved in HPCC primarily
because of the potential for HPCC technology to enhance their ability to carry
out agency missions.  NASA's projects, for example, are all linked to either
(1) design and simulation of aerospace vehicles or (2) earth and space sciences
research.  Rather than investing heavily in research to design new computer
architectures and build new systems, NASA concentrates on the use and
evaluation of HPCC systems in the context of its mission needs.  DOE similarly
emphasizes the role of being an early user of advanced systems and providing
feedback to the systems' developers, rather than attempting to develop new
system architectures on its own.  Both NASA and DOE have laboratories and
centers with extensive HPCC resources.  Much of their HPCC funding goes to
projects at these sites. 


   RECENT CHANGES IN PROGRAM DIRECTION
-------------------------------------------------------------------- Chapter 1:3

In February 1993, the new administration issued a document outlining its
strategy for investing in advanced technology.\2 In the document, the
administration rejected the traditional approach of limiting the federal
government's technology development spending to support of basic science and
mission-oriented research in the Department of Defense (DOD), NASA, and other
agencies.  The document stated that challenges facing the U.S.  were too
profound to rely on the government's investments in defense and space
technology to trickle down to the private sector.  Instead, it called for
direct support of private sector technology development efforts. 

In keeping with this new thinking, the administration sought to align the HPCC
program more closely with broader applications that could be developed and
commercialized in the private sector.  Specifically, HPCC was linked to the
development of a national information infrastructure (NII).  OSTP envisions the
NII, which is commonly referred to as the "information superhighway," as a
nationwide infrastructure of high performance computing hardware and massive
computer databases, all linked together by high-speed communications networks
and new software that allows trained users to access and use the information
contained therein. 

The HPCC program's technology support for the NII is contained in a new program
component added for fiscal year 1994, called Information Infrastructure
Technology and Applications (IITA).  The new component is intended to (1)
develop the technology base for the NII and (2) work with industry in using
this technology to develop and demonstrate new applications for the NII.  The
IITA component is also expected to broaden the market for HPCC technologies and
accelerate industry development of the NII. 


--------------------
\2 President William J.  Clinton and Vice President Albert Gore, Jr.,
Technology for America's Economic Growth, A New Direction to Build Economic
Strength, February 22, 1993. 


   HPCC'S TECHNICAL APPROACH
-------------------------------------------------------------------- Chapter 1:4

In addition to the new IITA component, the HPCC program includes efforts
undertaken in four other broad areas described below. 


      HIGH PERFORMANCE COMPUTING
      SYSTEMS
------------------------------------------------------------------ Chapter 1:4.1

The High Performance Computing Systems component concentrates on the
development of the underlying technology required to build scalable,\3 parallel
computer systems capable of sustaining trillions of operations per second on
large problems.  Most traditional computers have one computational processor,
and traditional computer development has focused on making this processor
faster and more efficient.  However, the potential for continued increases in
speed is reaching the limits imposed by the physical properties of the
materials used to build the processor.  Consequently, an entirely new kind of
computer design is needed if speed and performance improvements are to
continue. 

Computer scientists see development of parallel processing systems as the only
way to achieve the dramatic improvements in computer speed that will be needed
to address large, complex scientific problems.  Parallel processing means
breaking computational problems into many separate parts and having a large
number of processors tackle those parts simultaneously.  Greatly increased
processing speed is achieved largely through the sheer number of processors
operating simultaneously, rather than through any exceptional power in each
processor.  Massively parallel processing (MPP) refers to large machines that
include many cooperating processors.  Other approaches to parallel processing
include clustering large numbers of independent workstations together or
developing ways to link together a number of completely different computer
systems to address a single complex problem in parallel. 


--------------------
\3 Using the same basic architecture and system software, scalable machines
function effectively in configurations that range from a small number of
processors to a very large number--hundreds or even thousands--of processors. 


      ADVANCED SOFTWARE TECHNOLOGY AND
      ALGORITHMS
------------------------------------------------------------------ Chapter 1:4.2

The primary justification for developing increasingly more powerful parallel
computer systems is to address the large, complex scientific problems, commonly
known as the "grand challenges." The grand challenges are fundamental problems
in science and engineering that require significant increases in computational
capability to address, such as predicting global climate change or testing
advanced aircraft designs.  The Advanced Software Technology and Algorithms
component of the HPCC program targets software development to make MPP and
other high performance computer systems useful in addressing grand challenges. 

Radically new system software\4 and software tools are needed to operate MPP
and other parallel systems.  Most potential users have not yet adopted parallel
systems because of the high cost and risk of developing software for their
specific applications, and because system software for current parallel systems
is still rather primitive.  Major workshops on HPCC software convened in 1992
and 1993 agreed that greater focus on research to improve system software and
software tools is critical if the HPCC program is to succeed. 


--------------------
\4 System software is the collection of programs and data that make up and
relate to the operating system (for example, input/output routines,
command-line interpreters, and task scheduling and memory management routines). 


      NATIONAL RESEARCH AND EDUCATION
      NETWORK
------------------------------------------------------------------ Chapter 1:4.3

The National Research and Education Network segment of the program focuses on
the development of a national high-speed communications infrastructure to
enhance the ability of U.S.  researchers and educators to perform collaborative
research and education activities, regardless of their physical location or
available local computational resources.  This segment has two parts:  (1)
development of an interagency internetwork and (2) gigabit research and
development.  The interagency internetwork program will upgrade the networks of
participating agencies to higher speeds than are currently available and ensure
their interconnection.  The gigabit research and development program will
develop new high-speed communications technologies through basic research and
through experimentation with testbed networks located at various sites around
the country. 


      BASIC RESEARCH AND HUMAN
      RESOURCES
------------------------------------------------------------------ Chapter 1:4.4

The Basic Research and Human Resources segment supports long-term research by
individual investigators in scalable high performance computing.  It is also
intended to increase the pool of trained personnel by enhancing education and
training in HPCC.  Finally, this segment provides computing and communications
resources needed to support these research and education activities. 


      APPLICATIONS
------------------------------------------------------------------ Chapter 1:4.5

With the recent addition of the IITA component, the HPCC program now targets
two kinds of applications as ultimate beneficiaries of the technology being
developed in the program.  Program managers refer to these two groups of
applications as "grand challenges" and "national challenges."

Grand challenges, mentioned above, are aimed primarily at the scientific
research community.  National challenges, on the other hand, are defined in
HPCC program documentation as major societal needs that HPCC technology can
address, such as the civil infrastructure, digital libraries, education and
lifelong learning, energy management, the environment, health care,
manufacturing processes and products, national security, and public access to
government information.  While grand challenges address complex scientific
questions, national challenges involve making use of large stores of data and
information to enhance everyday activities.  The national challenges are an
identified subset of the wide range of potential applications that may be
developed for the NII. 


   OBJECTIVES, SCOPE, AND METHODOLOGY
-------------------------------------------------------------------- Chapter 1:5

In April 1993, the House Committee on Armed Services requested that we evaluate
the status of the HPCC program.  On the basis of subsequent discussions with
committee staff, our specific objectives were to assess (1) the effectiveness
of the program's management structure in setting goals and measuring progress
and (2) how extensively private industry has been involved in the planning and
execution of the program. 

To meet our objectives, we reviewed official HPCC program documentation of the
participating agencies and the NCO.  We also reviewed the administration's
statements regarding technology policy and the creation of the National
Information Infrastructure.  We discussed these issues with government,
academic officials, and private industry from a broad range of organizations. 

Specifically, with regard to the program's management, we interviewed
government officials at

Office of Science and Technology Policy, Executive Office of the President,
Washington, D.C.,

National Economic Council, Executive Office of the President, Washington, D.C.,

Office of Management and Budget, Executive Office of the President, Washington,
D.C.,

National Coordination Office for HPCC, National Library of Medicine, Bethesda,
Maryland,

ARPA, Computing Systems Technology office, Arlington, Virginia,

Department of Energy, Office of Energy Research, Gaithersburg, Maryland,

NASA, High Performance Computing and Communications Office, Washington, D.C.,

National Science Foundation, Directorate for Computer and Information Science
and Engineering, Washington, D.C.,

National Security Agency, Ft.  Meade, Maryland, and

National Institutes of Health, Bethesda, Maryland. 

We also interviewed officials from government laboratories, including

Oak Ridge National Laboratory, Oak Ridge, Tennessee,

Cornell Theory Center, Ithaca, New York,

National Center for Supercomputer Applications, Urbana- Champaign, Illinois,

Pittsburgh Supercomputing Center, Pittsburgh, Pennsylvania, and

San Diego Supercomputing Center, San Diego, California. 

We interviewed members of the academic community from

National Research Council, Computer Science and Telecommunications Board,
Washington, D.C.,

Syracuse University, Syracuse, New York,

California Institute of Technology, Pasadena, California,

Rice University, Houston, Texas,

University of Washington, Seattle, Washington,

Stanford University, Stanford, California,

University of California, Berkeley, California, and

University of Colorado, Boulder, Colorado. 

Regarding industry's participation in the program, we reviewed reports prepared
by industry associations and interviewed representatives of these associations,
including

Computing Research Association, Washington, D.C.,

EDUCOM, Washington, D.C.,

American Electronics Association, Washington, D.C.,

Information Technology Association of America, Washington, D.C., and

Computer Systems Policy Project, Washington, D.C. 

We also interviewed industry officials representing

Electronic Data Systems Corporation,

Eastman Kodak Company,

Microelectronics and Computer Technology Corporation,

Boeing Computer Services,

Visual Numerics, Inc.,

Tera Computer Company,

Schlumberger Well Services,

General Motors Research Corporation,

MasPar Computer Corporation,

Silicon Graphics, Inc.,

Sun Microsystems, Inc.,

Eli Lilly & Company,

Intel Corporation, and

Cray Research, Inc. 

A detailed audit of the funding of the HPCC program was beyond the scope of
this review.  Accordingly, we did not attempt to determine the appropriateness
of funding for any specific HPCC projects or the merits of proposals that have
not been funded.  However, we did collect budget information from each of the
six agencies included in the review in order to assess the program's management
processes for tracking and reporting how funds are spent. 

We conducted our review from May 1993 to June 1994, in accordance with
generally accepted government auditing standards.  The Assistant to the
President for Science and Technology provided written comments on a draft of
this report.  These comments are presented, along with our evaluation, in
appendix I. 


NEW PROGRAM DIRECTION COULD BENEFIT
FROM A MORE FOCUSED MANAGEMENT APPROACH
====================================================================== Chapter 2

The HPCC program is a loosely coordinated group of research and development
activities sponsored by a variety of federal agencies.  To date, the program's
broad technical goals have been driven by scientists' need for ever-increasing
computer power to address the grand challenges.  Now, however, the
administration is also counting on the HPCC program to help develop the new
technology that will be needed to make the NII successful and to give the
nation a competitive economic edge. 

In order to best ensure that it stays focused on achieving these more immediate
goals, the HPCC program could use more explicit management controls.  First,
the program will need to set more specific, measurable technical goals by
developing a prioritized technical agenda.  Such a document would serve as a
master program plan, identifying and prioritizing specific technical challenges
and establishing a framework for managing costs and evaluating results. 
Second, the program could make HPCC budget and expenditure information more
consistent and meaningful across participating agencies to improve public
visibility into program funding patterns. 


   CURRENT HPCC MANAGEMENT APPROACH
-------------------------------------------------------------------- Chapter 2:1

As discussed in chapter 1, the HPCC program involves 10 federal agencies that
have a wide variety of missions and approaches to research and development.  A
representative from the White House Office of Science and Technology (OSTP)
stated, and researchers we contacted agreed, that this diversity of management
approaches is a valuable asset in a research environment because it allows a
variety of technical approaches to be explored.  In addition, the major
participating agencies--ARPA, DOE, NASA, and NSF--had conducted successful
research and development programs related to HPCC for some years prior to the
establishment of the joint program.  As such, the OSTP representative stated
that designating a strong central manager for the HPCC program would not be
appropriate and that it would be disruptive to the programs to impose outside
control over them. 

Instead of taking a centralized approach, the OSTP representative said
participating agencies should be seen as members of a consortium, each pursuing
their own objectives but coordinating their efforts.  The agency program
managers are members of a committee called the High Performance Computing,
Communications, and Information Technology (HPCCIT) committee.  This committee,
which is chaired by the NCO, meets on a monthly basis to coordinate HPCC
activities.  A number of researchers told us that, to date, this arrangement
has worked reasonably well.  HPCC managers are generally given high marks by
researchers for sharing information and coordinating their activities.  Figure
2.1 shows the organization of the HPCC program. 

   Figure 2.1:  Organization of the
   HPCC Program

   (See figure in printed edition.)

The program originally operated under the assumption that the advances it
pioneers in high performance computing would eventually work their way down to
widespread use for everyday activities throughout the private sector.  Indeed,
much computer research funded by ARPA and DOD in the past for military
applications has been the foundation for technology widely used today in
personal computers and communications networks.  However, the administration
now argues that the challenges facing the U.S.  are too profound to rely on the
government's investments in defense and space technology to trickle down to the
private sector.  The administration intends the HPCC program to play a key role
in a more focused approach to stimulating commercial development and
application of new technologies. 

Measuring progress within the program remains an informal process.  In October
1992, OSTP established guidelines for the formal ongoing evaluation of federal
research programs.  Although the guidelines require that a program such as HPCC
submit a plan for continual and thorough evaluation of progress and outcomes,
no such plan has yet been prepared. 


   THE PROGRAM'S TECHNICAL AGENDA COULD
   BE BETTER DEFINED
-------------------------------------------------------------------- Chapter 2:2

Potential NII applications will require specific new technologies that the HPCC
program has not yet identified and prioritized as technical goals.  For
example, users of the NII will need to access and manipulate databases of
information that are much larger than can be handled efficiently by today's
systems.  Although some large database technology research is going on within
the HPCC program, no determination has yet been made about whether it is a
priority area that should be emphasized.  Outside commentators on the HPCC
program have proposed a range of specific technology areas, such as this, that
could be targeted as a way of accelerating development of the NII. 

Rather than targeting specific technology areas for accelerated development,
the HPCC program has pursued research in many different aspects of advanced
parallel computing.  The program has had two broad technical goals, which it
originally set out to achieve by 1996.  One is to gain a thousand-fold
improvement in useful computing capability and the other is to achieve a
hundredfold improvement in computer communications capability.\1 The program
has aimed to address the full spectrum of hardware, software, networking, and
training issues associated with developing this radically new breed of parallel
computers. 

Although much faster computers and networks are certainly a basic need,
particularly to enable scientists to address grand challenge problems, these
goals are all-encompassing and do not give enough technical focus to the
program.  Because they are so broad, controversy and confusion have sometimes
arisen as to what the "real" goals of the program are.  For example, university
and industry experts have observed that, in its original form, the program
appeared to be concentrating heavily on developing new hardware architectures,
with relatively little attention being paid to software issues, thus leaving
systems difficult to use.  More recently, the addition of the NII-oriented IITA
component has further broadened the technology spectrum to be addressed by
HPCC.  Both participants and outside observers have questioned the extent to
which the program is actually shifting its emphasis toward NII technology
issues, given that the level of funding for IITA projects to develop
applications in areas such as education and health care is minimal compared
with funding for hardware systems development. 

No official prioritization has yet been made.  The program's annual report to
the Congress describes ongoing work in a number of technical areas, but does
not prioritize among competing technical goals.  For example, the annual report
states that the five broad component areas of the program are considered
equally important.  Within the new IITA component, the document identifies a
range of technologies that will be needed for the NII, but does not prioritize
them or offer an overall strategy for developing them. 

An explicit technical agenda, identifying and prioritizing specific technology
challenges and establishing a framework of expected costs and results, could go
a long way toward better defining the program's direction.  This agenda could
also provide the needed management framework for focusing on technologies in
support of the NII.  Although it could take a variety of forms, an official
technical agenda would specify a target amount of resources to be invested in
each priority area and the major results that are expected.  Subject to
periodic review and adjustment, this document would clarify the program's goals
and objectives, focus efforts on critical areas, and serve as a baseline for
measuring program progress and results. 

One potential model for identifying and prioritizing technology challenges is a
draft prepared by the Computer Systems Policy Project, an affiliation of
American computer companies that have an interest in the national information
infrastructure.\2 The document identifies nine technology areas that will be
critical to the success of the NII.  For each of these nine areas, the document
lists a number of specific technologies that need to be researched and
developed and suggests which of these should receive priority attention. 


--------------------
\1 Computers of this type would be able to carry out a trillion or more
operations per second (one or more "teraops"), and communications networks
would be capable of transmitting data at a rate of a billion bits (one
"gigabit") per second. 

\2 Computer Systems Policy Project, Perspectives on the National Information
Infrastructure:  Accelerating the Development of NII Technologies, draft
version. 


   INCONSISTENT BUDGET INFORMATION HAS
   MADE TRACKING HPCC INVESTMENTS
   DIFFICULT
-------------------------------------------------------------------- Chapter 2:3

Budgets and expenditures for HPCC activities, both inside and outside the
program, have not been accounted for in a uniform and easily understood way. 
Accordingly, it is unclear how much money is actually being spent on advanced
computing and communications and on what projects. 

Spending for the formal HPCC program has closely followed its original plan of
expanding in even increments from a $500 million base program to approximately
$1.1 billion in its fifth year.  However, the program budget, which is often
cited publicly as a measure of the federal government's investment in HPCC,
actually offers little insight into how the federal government is investing in
total in HPCC research and development.  This is because participating agencies
have diverse research programs and equally diverse ways of identifying and
categorizing their HPCC spending.  There are no uniform guidelines for
determining what projects to include within the HPCC program or for
categorizing those projects within the five major components of the program. 


      BUDGET LEVELS DO NOT REFLECT ALL
      HPCC-RELATED FUNDING
------------------------------------------------------------------ Chapter 2:3.1

According to the official summary documents for the HPCC program that accompany
the President's budget request each year, nearly $3 billion has already been
spent on HPCC, and, beginning in fiscal year 1995, annual budgets will top $1
billion.  However, these figures do not reflect the total federal investment in
HPCC.  Several types of research and infrastructure activities are not
consistently included or excluded from the program. 

For example, preexisting government supercomputer centers have sometimes been
included in the HPCC program and sometimes not.  Four supercomputer centers
supported by NSF are included, as is the National Cancer Institute's center;
however, the supercomputer center at the National Center for Atmospheric
Research, also funded by NSF, is not included.  Similarly, NASA includes some
of its supercomputer facilities but not others.  In each case, program managers
have made their own judgments on what to include under HPCC since no
programwide guidelines were available. 

Advanced computer research that is not directly related to development of
scalable parallel computers is another area that is neither clearly within nor
clearly excluded from HPCC.  NSF includes research into advanced optical
computing, for example, whereas ARPA keeps its optical computing research
separate from HPCC.  NSF's HPCC program also supports fundamental research in
areas such as the theory of computing, software engineering, and the
theoretical aspects of computer systems, while ARPA funds this type of research
outside the HPCC program. 


      CATEGORIZATION BY PROGRAM
      COMPONENTS HAS NOT BEEN
      CONSISTENT
------------------------------------------------------------------ Chapter 2:3.2

HPCC program documentation uses five component categories to describe the types
of research and development that are funded within the program (these five
categories are defined in chapter 1).  Although this categorization could be
helpful in understanding how HPCC funds are spent, its value is diminished by
discrepancies in the way agencies categorize their official HPCC spending. 
Currently, no uniform method for categorizing projects is used.  Relying on the
personal judgment of HPCC managers and coordinators, participating agencies
group similar projects differently within the five program categories. 

For example, program documentation generally describes High Performance
Computing Systems as the hardware component of the program.  However, hardware
spending also shows up in the Advanced Software Technology and Algorithms and
Basic Research and Human Resources categories. 

The Basic Research and Human Resources component, in particular, overlaps all
the other categories, since program managers have to determine whether research
is "basic" and then categorize their projects accordingly.  Program managers
have listed a full spectrum of research activities under this component, from
research on architectures and systems, to software, algorithms, and
applications.  Because of these inconsistent classifications, it is difficult
to determine what areas HPCC is really emphasizing--developing hardware
platforms, writing systems software and tools, developing software
applications, or none of these--or how much effort is being expended on each. 


      NO EXPLICIT BUDGET GUIDELINES
      EXIST
------------------------------------------------------------------ Chapter 2:3.3

Explicit guidelines for preparing HPCC budgets across agencies, which do not
currently exist, would afford greater visibility into the overall federal
investment and would facilitate more informed assessments of whether
appropriate emphasis is being placed on areas that need greatest attention. 
Such guidelines should include new, more precise budget categories that would
provide visibility into how much is to be spent on operating supercomputer
centers, placement of computer systems, and other activities that support
researchers but may not be research per se. 

In April 1994, the NCO issued a document providing a detailed analysis of the
types of activities that each HPCC agency funds and how much is being spent for
them.\3 The new document is a step in the right direction in that it sets a
standard format for all participating agencies to use in presenting budget
information and presents more detailed information than has been publicly
available before.  However, the document does not resolve the discrepancies in
how various agencies account for their HPCC activities. 

In addition to increasing visibility into the government's investment, more
open and consistent reporting of HPCC funding could also broaden industry
support for the program, because the program's major interest areas and
priorities for funding would be clearer. 


--------------------
\3 FY 1995 Implementation Plan, National Coordination Office for HPCC, April 8,
1994. 


   CONCLUSIONS
-------------------------------------------------------------------- Chapter 2:4

While continuing to foster basic research to address scientists' need for
ever-increasing computer power to address grand challenge problems, the HPCC
program is also taking on the task of developing the specific technologies that
will be needed for the NII.  In order to be successful at that new task, the
program could benefit from a detailed technical agenda, identifying and
prioritizing the kinds of technologies it will develop in support of the NII. 
Such a document would better define the program's direction and also serve as a
baseline for measuring future progress. 

The budget information annually reported to Congress on HPCC does not provide
enough visibility into how much the government is investing in HPCC or what
kinds of research and other activities are being funded.  Much of the problem
is due to the fact that no precise guidelines exist for determining what
activities to include within the HPCC program.  Also, the program's five
component categories, while useful in describing the program generally, are not
helpful in revealing the specific kinds of activities that are being funded. 


   RECOMMENDATIONS
-------------------------------------------------------------------- Chapter 2:5

We recommend that the Director of OSTP direct the HPCC program managers, in
consultation with industry and academic representatives, to develop an explicit
HPCC technical agenda, delineating the program's overall strategy and setting
development priorities for specific technology areas.  This document should
specify target amounts of resources to be invested in each priority area and
the major results that are expected, so that it can be used as a baseline for
measuring progress and controlling costs. 

We also recommend that the Director of OSTP develop, in consultation with the
Office of Management and Budget and the Congress, detailed guidelines for
preparing HPCC budgets, including guidance on the types of activities to
include in the program and how they should be categorized.  OSTP may wish to
delegate this task to the NSTC Committee on Information and Communications. 


   AGENCY COMMENTS AND OUR EVALUATION
-------------------------------------------------------------------- Chapter 2:6

In his September 1994 comments, the Assistant to the President for Science and
Technology (Science Advisor) concurred with our findings that a more focused
management approach is appropriate, given the new direction of the HPCC
program.  He said that this more focused approach will include improved
consistency in preparation of HPCC budgets within participating agencies as
well as a more detailed and prioritized technical agenda to ensure that the
goals of the program are clearly defined and success is clearly measurable. 

The Science Advisor disagreed with what he perceived as our view that the
program be centrally managed and that it have a centrally controlled budget. 
However, we did not recommend centralizing the program's management or budget;
instead, we discussed the advantages of a coordinated approach as well as the
drawbacks of central management.  We agree that HPCC program goals can be met
within the framework of the existing program structure.  However, achieving and
sustaining the kind of targeted effort now envisioned for HPCC must begin with
the identification of specific technical goals and priorities.  These specific
goals and priorities, once established, can then form an objective framework
for making decisions about the type of activities to be funded within the
program and the amount of funding to be allocated for each. 

In July 1994, a committee of the National Research Council issued an interim
report on the HPCC program that raised concerns in many of the same areas that
we addressed.\4 The committee, whose study is still ongoing, said it would
continue to examine areas such as the potential for developing standard program
performance measures for HPCC and the need for greater budget consistency. 


--------------------
\4 Interim Report on the Status of the High Performance Computing and
Communications Initiative, National Research Council, July 1, 1994. 


GREATER INDUSTRY INVOLVEMENT COULD HELP
THE HPCC PROGRAM
====================================================================== Chapter 3

Close collaboration with industry is essential to ensure that the HPCC program
meets its goal of accelerating the development and widespread use of HPCC
technologies.  While industry has been extensively involved in the actual
execution of HPCC projects, as the program moves forward it would benefit from
partnerships with key industries that could capitalize on HPCC technologies to
create new products and services for the NII. 

Representatives from a variety of companies with a potential interest in HPCC
told us they remain uninvolved in the program for several important reasons. 
They expressed the belief that the program does not address their needs and
interests, largely because HPCC managers have not solicited their input in
program planning.  Also, the NCO, which was established in part to foster
industry participation, has not provided industry representatives with needed
information or responded to industry initiatives to improve communications
between the program and potential industry participants.  Given that the
administration sees the HPCC program as playing an important role in developing
key technologies for the NII, HPCC managers must more effectively promote
industry participation. 


   ACHIEVING NII-RELATED GOALS REQUIRES
   GOVERNMENT RESEARCHERS TO
   COLLABORATE CLOSELY WITH INDUSTRY
-------------------------------------------------------------------- Chapter 3:1

Since the program's inception, HPCC program documentation has emphasized that
industry participation is critical to meeting the program's goals of
accelerating the development and widespread application of high performance
computing and networks.  Now, industry's collaborative role has become even
more important in the context of HPCC's new role of supporting development of
the NII. 

Specifically, the HPCC program is now committed to helping the private sector
develop new technologies, including applications and services, that will
maximize the value of the NII to a broad base of users.  These applications
include remote medical diagnosis by specialists and experts anywhere in the
nation; the delivery and use of environmental information for a broad range of
users, such as agriculture workers and truckers; and enhanced educational
opportunities in which students could perform science experiments in
collaboration with scientists at the national laboratories or visit museums and
research centers without leaving their classrooms.  In each case, it is
envisioned that these applications will be developed by the private sector,
with some level of government support.  One goal of government collaboration
will be to help ensure that issues of accessibility, security, and reliability
are addressed. 


   INDUSTRY HAS FOCUSED ON EXECUTION
   OVER PROGRAM PLANNING
-------------------------------------------------------------------- Chapter 3:2

Industry involvement in the actual execution of the HPCC program has been
extensive.  At ARPA alone, for example, 43 percent of the HPCC budget goes to
companies that have successfully responded to ARPA's requests for research
proposals in specific technological areas.  DOE also has established
cooperative agreements with numerous partners from industry.  Nevertheless,
HPCC managers have generally not involved industry in planning the HPCC
program.  At the governmentwide level, a mechanism for obtaining nonfederal
advice and evaluation was mandated by the High Performance Computing Act, which
directed the President to establish an advisory committee including
representatives from industry.  According to OSTP officials, the administration
is working to get the advisory committee appointed, although concerns about
potential conflict-of-interest have slowed the effort. 

Many HPCC agencies have their own advisory committees that review their HPCC
programs.  These committees have been helpful in planning effective agency
programs.  A case in point is NASA's program, which was reviewed in 1993 by a
NASA Advisory Council Task Force.  The task force reported that the priorities
in the agency's HPCC plan did not address the research problems that the
aerospace industry considered most critical.  NASA responded by soliciting
direct industry involvement in reworking its program plan for aerospace. 
Aerospace industry representatives told us they are encouraged that a revised
plan will more fully reflect their interests and concerns. 

The NCO, which was established in part to serve as a point of entry for
industry into the program, disseminates general information about the program
as well as funding opportunities.  The NCO recently made this information
available electronically over the Internet.  In addition, the NCO has been
involved in numerous liaison activities with industry, academia, and the
public.  These activities have included meetings, workshops, and conferences. 
The NCO has also allowed groups of industry representatives to attend certain
designated portions of the HPCC program managers' regular meetings and give
brief presentations of their views. 

Industry representatives whom we contacted agreed that all of these activities
are valuable.  However, they seek greater opportunities for close collaboration
between government and industry in planning program direction.  They have
proposed that the NCO cooperate in arranging for the HPCC program to
participate officially in symposia, in order for industry and academic
representatives to meet with program managers to air their views on the
direction and priorities of the program.  They emphasize that these meetings
should provide for a full discussion and consideration of issues of importance
to industry, such as how best to invest limited resources.  The NCO could
exercise this function until a permanent advisory committee, which will
maintain a more substantial, ongoing dialogue with program management, is
appointed. 


   GREATER INDUSTRY INVOLVEMENT DEPENDS
   ON INCREASING SOFTWARE EMPHASIS
-------------------------------------------------------------------- Chapter 3:3

A major roadblock to broader industry utilization and commercialization of high
performance computing technologies is the lack of software and software
development tools to take advantage of the power of high performance computers. 
Currently, only a limited range of applications software is available, and
development tools, which are needed to write new applications software, are
primitive.  Moreover, a lack of standards discourages industry from investing
in software development projects that may have a limited market.  A greater
emphasis by the HPCC program on software could reduce some of the risks for
potential industry participants and increase their involvement. 


      EXISTING SOFTWARE APPLICATIONS
      AND TOOLS ARE INADEQUATE
------------------------------------------------------------------ Chapter 3:3.1

HPCC so far has focused on the grand challenges as target applications.  While
the grand challenges are important scientific problems, they involve only small
communities of scientists working in specialized areas.  For example,
applications developed in NASA's HPCC program are targeted at aerospace
engineers designing and simulating new aircraft.  Earth and environmental
scientists, likewise, will profit from various HPCC projects supported by NASA,
NSF, DOE, and ARPA.  As valuable as these lines of effort are, they do not
directly address broad areas where HPCC technology can benefit the NII, and
industry tends to view them as offering little opportunity for
commercialization. 

One of the most important industry applications of HPCC on the NII will be
information processing and management.  A core set of generic software for
processing, storing, searching, and retrieving multiple data types from very
large databases would have a broad range of commercial applications, ranging
from health care to banking.  For example, software for handling databases of
imagery would enable applications as diverse as remote medical consultations or
law enforcement. 

Software development tools, which would make it easier for software companies
to design and develop new applications, might offer a particularly good
opportunity to leverage government investment in HPCC.  A series of reports by
groups of HPCC researchers has identified and prioritized the tools that would
be needed to facilitate the development of a broader range of applications
software.  These include debugging tools, memory management tools, and
performance analysis tools, all of which would help to create a more productive
software environment. 

The HPCC program already supports some research in these areas.  However, by
establishing software development as a priority and devoting more resources to
it, HPCC would encourage industry to invest in the development of a wide range
of specialized NII applications. 


      STANDARDS-SETTING ACTIVITIES NEED
      GREATER SUPPORT
------------------------------------------------------------------ Chapter 3:3.2

Developers have identified the lack of standards as an impediment to more
intensive commercial development of HPCC applications software.  Agreement on
standards would permit commercial software developers to build programs that
work on a variety of high performance computers, rather than on only one
specific hardware system, which may or may not do well in the marketplace. 
Broadening the base of computers on which the software will run would expand
its potential commercial market, thereby allowing developers to put a much
greater effort into building applications software. 

However, setting standards is a difficult process, requiring a great deal of
interaction over time within the HPCC community.  Industry representatives
agree that the government should not set standards.  Industry, they believe,
must lead this effort.  Nevertheless, the government can play a practical role
in supporting standards-setting efforts. 

The HPCC program already provides funding for several standards- setting
activities.  For example, several agencies support a project to establish a
standard HPCC version of the Fortran programming language.  However, industry
representatives have urged greater government support for standards-setting
activities in order to stimulate commercial software development. 
Specifically, the HPCC program could fund more workshops where government,
academia, and industry can come together to discuss and collaborate on emerging
standards.  The program could also provide more direct support for researchers
to work with industry on evaluating potential standards. 


   CONCLUSIONS
-------------------------------------------------------------------- Chapter 3:4

It is widely recognized that the HPCC program needs a standing advisory
committee that includes representatives from a broad range of potential
industry participants.  Such a committee would provide the mechanism to sustain
an ongoing dialogue between the program and industry.  However, in addition to
establishing this committee, program officials can take additional steps to
promote industry involvement, through cosponsoring symposia with industry and
involving industry representatives in the program planning process, in order to
forge a true partnership between government and industry. 


   RECOMMENDATIONS
-------------------------------------------------------------------- Chapter 3:5

We recommend that the Director of OSTP (1) take steps to expedite the
appointment of an advisory committee whose membership includes representatives
from a wide range of industries, and (2) delegate to the NCO the role of
sponsoring symposia where industry can meet with program officials and academia
to help define the research priorities of the program. 

We also recommend that OSTP direct the Director of the NCO to take additional
steps to promote industry participation, including involving industry
representatives in the program planning process, and providing greater support
for software development and standards-setting activities to make it easier for
industry to develop applications for deployment on the NII. 


   AGENCY COMMENTS AND OUR EVALUATION
-------------------------------------------------------------------- Chapter 3:6

In his formal comments, the President's Science Advisor strongly concurred with
the recommendation that a private sector advisory committee be established and
noted that OSTP was taking the initial steps to do so.  The Science Advisor did
not comment on our recommendation that the NCO sponsor symposia involving
industry, academia, and HPCC program managers. 

In preliminary discussions on a draft of the report, HPCC program managers
maintained that the program has already implemented our recommendation to place
greater emphasis on developing software tools and sponsoring standards-setting
activities, as documented in the fiscal year 1995 Implementation Plan.  We,
however, do not agree that a significant shift in emphasis has yet occurred. 
While the implementation plan recognizes that greater focus on software tools
will be required to encourage industry involvement in developing applications,
a small percentage of the budget for the advanced software technology and
applications component is allocated to this area.  We believe that the program
could better leverage federal funding by devoting more resources to activities
that would make it easier for private industry to develop a broader range of
applications. 

In its interim report, the National Research Council's HPCC study committee
expressed concerns similar to ours.  The committee recommended that an HPCC
Advisory Council be appointed immediately to provide broad-based, active input
to the HPCC program from industry and academia as well as government.  The
committee also expressed concerns about the need for software development to
catch up with advances that have been made in HPCC hardware development. 




(See figure in printed edition.)Appendix I
COMMENTS FROM THE ASSISTANT TO THE
PRESIDENT FOR SCIENCE AND TECHNOLOGY
====================================================================== Chapter 3

See comment 1. 



(See figure in printed edition.)

See comment 2. 



(See figure in printed edition.)


The following are GAO's comments on the letter from the Assistant to the
President for Science and Technology dated September 20, 1994. 


   GAO COMMENTS
-------------------------------------------------------------------- Chapter 3:7

1. We discussed a draft of the report with the High Performance Computing,
Communications, and Information Technology (HPCCIT) committee, which is
composed of representatives from each of the agencies participating in HPCC. 
The HPCCIT committee also provided us with preliminary written comments. 

2. This issue is discussed in the "Agency Comments and Our Evaluation" section
of chapter 2. 


MAJOR CONTRIBUTORS TO THIS REPORT
==================================================================== Appendix II


   ACCOUNTING AND INFORMATION
   MANAGEMENT DIVISION, WASHINGTON,
   D.C. 
------------------------------------------------------------------ Appendix II:1

Franklin W.  Deffer, Assistant Director
John A.  deFerrari, Assistant Director
Elizabeth L.  Johnston, Senior Evaluator
Shane D.  Hartzler, Reports Analyst

RELATED GAO PRODUCTS

Advanced Research Projects Agency Should Do More to Foster Program Goals
(GAO/IMTEC-93-24, May 17, 1993). 

Industry Uses of Supercomputers and High-Speed Networks (GAO/IMTEC-91-58, July
30, 1991). 


*** End of document. ***