.

DOCUMENT RESUME

ED 373 784 IR 055 139

TITLE High Performance Computing and Communications:Technology for the National InformationInfrastructure. Supplement to the President's FiscalYear 1995 Budget.

INSTITUTION Office of Science and Technology Policy, Washington,DC. National Science and Technology Council.

PUB DATE [94]

NOTE 70p.

PUB TYPE Reports Descriptive (141)

EDRS PRICE MF01/PC03 Plus Postage.DESCRIPTORS Computer Networks; Computer Software Development;

Federal Programs; Futures (of Society); Glossaries;Information Networks; *Information Technology;Research and Development; *Technological Advancement;*Telecommunicat ons

IDENTIFIERS *High Performance Computing; *National InformationInfrastructure; National Research and EducationNetwork

ABSTRACTThe federal High Performance Computing and

Communications (HPCC) program was created to accelerate thedevelopment of future generations of high performance computers andnetworks and the use of these resources in the federal government andthroughout the American economy. This report describes the HPCCprogram which is developing computing, communications, and softwaretechnologies for the 21st Century. The HPCC program will provide keyparts of the technological foundation for the National InformationInfrastructure. The program is organized into the following fivecomponents: High Performance Computing Systems (HPCS); NationalResearch and Education Network (NREN); Advanced Software Technologyand Algorithms (ASIA); Information Infrastructure Technology andApplications (IITA); and Basic Research and Human Resources (BRHR).The report is divided into four sections: executive summary; programaccomplishments and plans; high performance living today andtomorrow; and the HPCC program in summary. A glossary of terms and alist of contacts are included. (JLB)

Reproductions supplied by EDRS are the best that can be madefrom the original document.

12_4"' "s . . .

(.1

3) A,

U ti DEPARTMENT OF EDUCATIONOlhce et Educational Research and improvementEDUCATIONAL RESOURCES INFORMATION

CENTER (ERIC)

O Thni docunnl has been reproduced asrercrnyd from the person or organizationonginating it

O Minor Changes have Wen mad, to improvereproduction Quality

Points ot vein, Or opinions stated in this docu-mein! do not necessarily represent othcilOERI posillon or policy

e

BEST COPY AVAILAR F' ,

. v..* ...1)

Cover images:

1. Predicted (yellow) and observed (orange)track of Hurricane Emily

2. Simulation of the behavior of materials at thefundamental atomic scale

3. High precision manufacturing of a coppermirror for a laser welder

4. Blood streaming through the aortic valve in acomputer model of the heart

5. Father and daughter at a High PerformanceComputing Research Center

jBEST COPY AVAILABLE

HIGH PERFORMANCECOMPUTING AND

COMMUNICATIONSTECHNOLOGY FOR THE NATIONAL INFORMATION INFRASTRUCTURE

A Report by the Committee on Information and Communication

National Science and Technology Council

4

EXECUTIVE OFFICE OF THE PRESIDENTOFFICE OF SCIENCE AND TECHNOLOGY POLICY

WASHINGTON, D.C. 20500

MEMBERS OF CONGRESS:

I am pleased to forward with this letter "High Performance Computing and Communications:Technology for the National Information Infrastructure" prepared by the Committee onInformation and Communication (CIC) of the National Science and Technology Council(NSTC). This report, which supplements the President's Fiscal Year 1995 Budget, describesthe High Performance Computing and Communications (HPCC) Program, which prior to thecreation of the NSTC, was coordinated by the Committee on Physical, Mathematical, andEngineering Sciences (CPMES) within the Federal Coordinating Council on Science,Engineering, and Technology (FCCSET).

The interagency HPCC Program is developing computing, communications, and softwaretechnologies for the 21st century. It is fully supportive of and coordinated with theAdministration's National Information Infrastructure (NII) Initiative, which was described inthe NII Agenel. for Action released on September 15, 1993. The Program will provide keyparts of the technological foundation for the NII and develop and demonstrate selected"National Challenge" applications. National Challenges are major societal needs thatcomputing and communications technology can help address in key areas such as health care,manufacturing, digital libraries, education and lifelong learning, electronic commerce, energymanagement, the environment, national security, and government services.

NII technologies are critically interwoven with and dependent upon high performancecomputing and communications capabilities and the software needed to address scientific andengineering "Grand Challenges" such as forecasting the weather, building safer and moreenergy efficient aircraft, designing life saving drugs, and understanding how galaxies areformed. The Program has accelerated the development of these technologies by supportingthe researchers who create and apply them and the educators who teach others how to usethem.

In formulating its research and development agenda, the HPCCIT Subcommittee and itsparticipating agencies have worked closely with industry and academia. These keystakeholders have been providing informal input to the Program, and we are currentlyformalizing the process for ongoing private sector participation in conjunction with the newframework of the NSTC. Donald A. B. Lindberg, Chair of the HPCCIT Subcommittee.other members of the Subcommittee. their associates, and staff are to be commended fortheir efforts on this report and on the Program itself.

. Gi nsAssista ) the President

for Science and Technology

Table of Contents

I. Executive Summary 1

H. Program Accomplishments and Plans

1. Networking 7

I.I. The Internet 7

1.1. The Interagency Internet 7

1.3. Gigabit Speed Networking R&D 9

1.4. Network Security 10

2. High Performance Computing Systems 10

3. High Performance Computing Research Centers (HPCRCs)4. Software 14

4.1. Systems Software and Software Tools 15

4.1. Scientific Computation Techniques 16

4.3. Grand Challenge Applications 16

Aircraft 17

Computer Science 17

Energy 18

Environmental Monitoring and Prediction 18

Molecular Biology and Biomedical Imaging 20Product Design and Process Optimization 2ISpace Science 12

5. Technologies for the National Information Infrastructure (NII)5.1. Information Infrastructure Services5.2. Systems Development and Support Environments5.3. Intelligent Interfaces6. National Challenges

Digital Libraries 25

Crisis and Emergency Management 25Education and Lifelong LearningElectronic Commerce (EC) 26

Energy Management 26

Environmental Monitoring and Waste Minimization 26Health Care 28Manufacturing Processes and Products 29

Public Access to Government Information7. Basic Research 308. Training and Education 319. HPCC Program Management 3210. Information about the HPCC Program 33

III. High Performance Living Today and Tomorrow 35

Crisis Maragement: Earthquake Relief in the Year 2000 36Education and Lifelong Learning 37

Education and the NII Circa 2000 37

Lifelong Learning 37Electronic Commerce 38

Electronic Banking 38Electronic Brokering 38

Buying a Car in the Year 2000 39Increased Productivity through Improved Environmental Data 40Environmental Monitoring in the Year 2000 41

Delivering Health Care to Remote Areas 42

IV. The HPCC Program in Summary

HPCC Program Goals 45HPCC Agencies 45HPCC Program Strategies 46Overview of the Five HPCC Program Components 47Evaluation Criteria for Agency Participation in the HPCC Program 48Agency Responsibilities 49Agency Budgets by HPCC Program Components 50HPCCIT Reporting Structure and Subcommittee Roster 51

V. Glossary 53

VI. Contacts 57

Executive Summary

The Federal High Performance Computing andCommunications (HPCC) Program was createdto accelerate the development of future genera-tions of high performance computers and net-works and the use of these resources in theFederal government and throughout theAmerican economy.

In the early 1980s American scientists, engi-neers. and leaders in government and industryrecognized that advanced computer and commu-nications technologies could provide vast bene-fits throughout not just the research communitybut the entire U.S. economy. Senior govern-ment. industry, and academic scientists and man-agers initiated and are implementing the HPCCProgram to extend U.S. leadership in these tech-nologies and to apply them to areas of profoundimpact on and interest to the American people.The National High-Performance ComputingProgram was formally established following thepassage of the High Performance ComputingAct of 1991 (Public Law 102-194). introducedby then Senator Gore.

The scalable high performance computing sys-tems. advanced high speed .computer communi-cations networks, and advanced software beingdeveloped in the HPCC Program are necessaryfor science and engineering and will contributecritical components of the National InformationInfrastructure (N11). This infrastructure is essen-tial to our national competitiveness. It willenable us to build digital libraries, enhance edu-cation and lifelong learning. manage our energyresources better. monitor and protect the envi-ronment. and improve health care. manufactur-ing. national security. and public access to gov-ernment information.

The Program is planned. funded. and executedthrough the close cooperation of Federal agen-cies and laboratories, private industry, andacademia. These efforts are directed towardensuring that io the greatest extent possible theProgram meets the needs cl all communitiesinvolved and that its results arc brought into the

1

research and educational communities and intothe commercial marketplace as effectively aspossible.

The Program is well on its way toward meetingits original goals:

More than a dozen high performance com-puting research centers are in operationnationwide. More than 100 scalable highperformance systems are in operation atthese centers. These include large scale par-allel systems. vector parallel computing sys-tems. hybrid systems. workstations andworkstation clusters. and networked hetero-geneous systems. The largest of these sys-tems now provides more than 50 gigaflops(billions of floating point operations per sec-ond) performance on large problems. TheHPCC Program's 1996 goal of developinghigh performance systems capable of sus-tained teraflops (trillions of floating pointoperations per second) performance is wellon the way to being met.

The HPCC-funded communications net-works, part of the Internet. continue to expe-rience phenomenal growth in size, speed,and number of users. These networks enableresearchers to access high performance com-puters and advanced scientific instrumentseasily. and have begun to allow indepen-dence of geographical location.

One illustration of the global reach of HPCCtechnology is that the Internet now extendsacross the country and throughout much ofthe world. The Internet links more than twomillion computers. more than 1 c.000 net-works in the U.S. and more t'aan o.0(X) out-side the U.S., and 1.0(X) 4-year colleges anduniversities, 1(X) community colleges. 1.000high schools, and 300 academic libraries inthe U.S.

More than half a dozen gigabit testheds con-duct research in high capacity networks.

.-4.-2'qt1111=11111111

These net ork testheds connect 24 sites.including many of the high performancecomputing research centers. Seven Federalagencies, 18 telecommunications carriers, 12universities, and two state supercomputercenters participate. The testheds developtechnologies to handle the NIrs increaseddemand for computer communications, alongwith greater accessibility. interoperability,a..d security. The Program goal of demon-strating gigabit (billions of hits) per secondtransmission speeds by 1996 is well on theway to being met.

Teams of researchers are using scalable sys-tems to discover new knowledge and demon-strate new capandities that were not possiblewith earlier technologies. These researchersare addressing the "Grand Challenges." fun-damepml problems in science and engineer-ing with broad economic and scientificimpact whose solutions can he advanced byapplying high performance computing tech-niques and resources. Many of these chal-lenges are associated with HPCC agencymissions. Agencies are increasingly workingwith U.S. industry to use their GrandChallenge applications software and developnew software that will improve commercialand industrial competitiveness.

Users of the new scalable computing systemswith their high performance on large prob-lems and larger memories arc able to addressmore complex. more realistic problems. Weare beginning to understand our world betterand to improve our lives:

Improved modeling of the Earth and itsatmosphere has sharpened our ability topredict the movement and characteristicsof storms and other forms of severe weath-er. With improved forecasts, there will bemore lives saved and an overall positiveeconomic impact due to reduced propertyloss and evacuation of smaller endangeredareas along the coast and elsewhere.

New air and Water quality models areenabling improed environmental decisionmak'ng.

2

.1,

Improvements in the design and manufac-ture of goods are yielding better products.Both the production processes and prod-ucts such as cars and airplanes are becom-ing more energy efficient.

We are learning more about how thehuman body functions and are improvingour ability to diagnose and treat diseases.

-I The thousands of researchers who developfundamental HPCC and NII technologies andapplications form the vanguard that hardwareand software vendors rely upon to promotethe use of high performance computers andcommunications throughout the U.S. econo-my. Hundreds of teachers and thousands ofstudents access HPCC resources, and theProgram conducts hundreds of trainingevents for thousands of trainees. Dozens ofsmall systems have been installed at collegesand universities. The goal of these efforts isto train a nation of knowledgeable users andthereby fully incorporate HPCC and Niltechnologies and applications into the U.S.economy.

In his State of the Union speech on January 24.1994, President Clinton called for connectingevery classroom, library. clinic. and hospital inAmerica into a "national information superhigh-way" by the year 2000. He said. "Instant accessto information will increase productivity. it willhelp educate our children. It will provide bettermedical care. It will create jobs." The HPCCProgram is helping to fulfill this vision by devel-oping much of the underlying technology for theNII, enabling the development of "NationalChallenge- applications, and demonstrating avariety of pilot applications. NationalChallenges are fundamental applications thathave broad and direct impact on the Nation'scompetitiveness and the well-being of its citi-/ens. and that can benefit from the application ofHPCC technologies and resources. The includecrisis and emergency management. digitallibraries, education and lifelong learning, elec-tronic commerce, energy demand and supplmanagement, environmental monitoring andwaste minimization. health care, design of man-

ufacturing processes and products, and publicaccess to government information. Some of theearly National Challenge applications, such aselectronic commerce, will mature to become ser-vices in the NIL enabling a wider range of futureNational Challenge applications.

Pending Congressional legislation introduced in'993 would formally expand the Program toinclude such responsibilities.

HPCC and the NII are tightly interwoven. Whilethe Federal agencies that participate in theHPCC Program are working to enable the NII, itis the private sector that will deploy it. Manypolicy and regulatory obstacles will be clearedthrough the workings of the interagencyInformation Infrastructure Task Force.

The capabilities being developed under theHPCC Program provide much of the technologybase critically needed for the NationalInformation Infrastructure:

J Scalable computing technologies providethe foundation for the computing systemsneeded by the NII. For example:

The microprocessors at the heart oftoday's most powerful scalable parallelcomputers appear in relatively inexpensivedesktop workstations and personal com-puters today and will he found in theinformation appliances in the homes oftomorrow.

The same technology that provides highspeed connectivity between computer pro-

National Information Infrastructure Layers

III

.

.

Electronic Transactions, . .

Data Interchange

Multimedia ObjectsCollaboration Support .

Resource Discovery,,,s

: s .

In FY 1994 the HPCC Program began expanding its technical scope to include enabling technologies toaccelerate the development of a National Information Infrastructure. The fundamentally advantageoustechnical properties of scalable high performance computing and communications make HPCC technolo-gies critical for the National Information Infrastructure.

3

BEST COPY AVAILABLE

::::Wiiiii1111111111111111111111111111111111,

censors and memory can he used to buildswitches for the NII's high speed net-works.

The same "client/server" technology usedto supply data from remote computers toGrand Challenge computations can beused to disseminate information toAmerican households.

NII applications such as the entertainmentindustry's "video on demand" will soonuse the scalable parallel computing andmass storage systems that have long beenpart of high performance computingresearch centers.

The technology for the high bandwidth net-works needed by Grand Challengeresearchers to move data between computersand to distribute information to users of theNII is being developed at the gigabittestheds.

-IThe software technologies, including operat-ing systems and software development envi-ronments, being developed under HPCChave widespread application to general-pur-pose systems, not just those identified as"numerically intensive."

Many National Challenge applicationsdepend upon computationally intensiveGrand Challenge applications. For example:

Improving health care requires biomedicalresearch (such as improved moleculardesign of drugs).

Improved environmental monitoringrequires computationally intensive envi-ronmental models.

Effective response to natural and man-made disasters depends on weather fore-casting and environmental models.

Developing advanced methods for productand process design and manufacturingrequires numerically-intensive prototypingof those products and processes.

4

The "information" in the NationalInformation Infrastructure will include thedata used by and the results from GrandChallenge research.

The NII places additional demands on HPCCresearch and development. Sca lability issuesmust be addressed the technologies used toserve thousands of researchers and educatorsmust be scaled up to serve millions of NII users.With the larger number of NH users and theirgreater diversity of interests and skills, moreubiquitous alternative "on-ramps" to the emerg-ing high bandwidth networks of the NII must beexamined. For example, at the Monte Vista HighSchool in Cupertino, CA, Vice President Gorevisited a class of students who were directlyaccessing the Internet over the local cable TVsystem. This technology was developed underHPCC sponsorship.

New services are being developed to supportmultimedia applications, digital libraries tech-nologies, appropriate p'vacy and security pro-tection, and increasingly higher levels of perfor-mance. Software development environmentsand building blocks for interfacing humans andcomputers must he created. These new toolswill enable application developers to constructcomplex, large-scale, network-based, user-friendly, and information-intensive applications.These technologies provide the foundation uponwhich National Challenge applications will bedeveloped.

The HPCC Program reports to the Director ofthe Office of Science and Technology Policy.At his direction, budget oversight is provided bythe National Science and Technology Councilthrough its Committee on Information andCommunication. The Program is managed bythe High Performance Computing,Communications, and Information TechnologySubcommittee. The National CoordinationOffice (NCO) for High Performance Computingand Communications completed its first year ofoperation in September 1993. During that year.it held more than 50 meetings witb representa-

rives from the U.S. Congress. state and localizovernments. foreign governments, industry, and

universities.

HPCC Program agencies work closely withother government agencies. industry. andik.ademia in developing. supporting, and usingHPCC and other NII-associated technologies.The NCO and HPCC Program agencies wort:with members of the Information InfrastructureTask Force in addressing NIL policy issues, andwits other Federal agencies involved in helpingschools, libraries, and medical facilities becomepart of the NII. Industrial. academic, and profes-sional societies also provide critical analyses ofthe HPCC Program through conferences, work-shops. and reports. Through these efforts,Program goals and accomplishments are betterunderstood and Program planning and manage-ment are strengthened.

The 1(1 agencies that participate in the HPCCProgram are:

ARPANSFDOENASA

NIHNSANIST

NOAA

EPAEl)

Advanced Research Projects AgencyNational Science FoundationDepartment of EnergyNational Aero"'uttics and SpaceAdministrationNational Institutes of HealthNational Security AgencyNational Institute of Standards andTechnologyNational Oceanic and AtmosphericAdministrationEnvironmental Protection AgencyDepartment of Education

Through coordinated planning and research anddevelopment, these agencies are developing anintegrated infrastructure of HPCC and NII tech-nologies. No individual agency has either themission or the expertise to develop all compo-nents of the infrastructure. but each plays a nec-essary and unique role in the overall program.

The HPCC Program is organized into five com-ponents with the following key aspects:

5

High Performance Computing Systems(HPCS)

-I Accelerated development of scalable com-puting systems. with associated software,including networks of heterogeneous sys-tems ranging from affordable workstations tolarge scale high performance systems

-I Technologies to enable the use of advancedcomponent. packaging, mass storage, andcommunications technologies for the designof large scale parallel computing systems

National Research and Education Network(NREN)

J Broadened network connectivity of theresearch and education communities to highNrformance computing and researchresources

J Accelerated development and deployment ofnetworking technologies

Advanced Software Technology andAlgorithms (ASTA)

Prototype solutions to Grand Challengeproblems

Improved algorithms. software technologies,and software tools for more efficient use ofscalable computing systems

Deployment of advanced high performancecomputing systems

Information Infrastructure Technology andApplications (DTA)

Prototype solutions to National Challengeproblems using HPCC enabling technologies

Accelerated development and deployment of

NIL enabling technologies

........-....11000.11

Basic Research and Human Resources(BRHR)

-I Support for research, training, and educationin computer science, computer engineering.and computational science, and infrastruc-ture enh..,:,..ement through the addition ofHPCC resources

The total FY 1994 HPCC Program budget for 10participating agencies is S93% million. For FY1905, the proposed HPCC Program budget fornine agencies is S1.155 billion, representing a 23percent increase over the appropriated FY 1994le',el.

6la

HPCC Program Accomplishments andPlans

1. Networking

The HPCC Program provides network connec-tivity among advanced computing resources. sci-entific instruments, and members of the researchand education communities. The Program hassuccessfully accommodated t'-.e phenomenalgrowth in the number of network users and theirdemands for significantly higher and everinc -easing speeds while maintaining operationalstability. R&D in advanced networking tech-nologies is guiding the development of a com-mercial communications infrastructure for theNation. The development and deployment ofthis new technology is jointly funded and con-ducted by the HPCC Program. state and localgovernments, the computer and telecominunica-tions industries, and academia.

1.1. The Internet

One illustration of the global reach of HPCCtechnologies is that the Internet now extendsacross the country and around much of theworld. Initially the domain of government sci-entists and U.S. academics, by the beginning ofFY 1994:

J Almost t..vo million computers were accessi-ble over '.he Internet.

J More than 15.000 regional. state. and localU.S. networks and 6.300 foreign networks inapproximately 100 countries were part of theInternet.

J Nearly 1.000 4-year colleges and universi-ties, 100 community colleges, 1.000 U.S.high schools, and 300 academic libraries inthe U.S. were connected.

The HPCC Program's Internet investment pri-marily supports the high speed "backbone" net-

works linking Federally-funded high perfor-mance computing centers.

1.2. The Interagency Internet

The Interagency Internet. that portion of theInternet funded by HPCC, is a system of' value-added services carried on the Nation's existingtelecommunications infrastructure for use in fed-erally-funded research and education. Its three-level architecture consists of high speed back-bone networks (such as NSFNET) that link mid-level or regional networks, which in turn connectnetworks at individual institutions. At the begin-ning of the HPCC Program in FY 1992, most ofthe backbones were running at T1 speeds (1.5Mb/s megabits per second or millions of bitsper second), and international connections hadbeen established. Peak monthly traffic onNSFNET had reached 10 billion packets (ofwidely varying size). In FY 1992. NSFNETspeed was upgraded to T3 (45 Mh/s) and NSFmade awards to industry for network registra-tion, information, and database services and for aClearinghouse for Networked InformationDiscovery and Retrieval. By the beginning ofFY 1994:

J Peak monthly NSFNET traffic reached 30billion packets.

J DOE established six ATM (asynchronoustransfer mode) testheds to evaluate differentapproaches for integrating this technologybetween wide area and local area networks.

-I NASA provided T3 services to two of itsGrand Challenge Centers (Ames ResearchCenter and Goddard Space Flight Center)through direct connection to NSFNET. Inaddition. service to several remote investiga-tors was upgraded to 'F1 data rates.

7

et

:vals

a NASA launched its AdvancedCommunications Technology Satellite(ACTS).

41+408_,;

Advanced Communications Technology Satellite(ACTS) deployed by the Space Shuttle.

a NIH and NSF funded 15 MedicalConnections grants for academic medicalcenters and consortia to connect to theInternet.

a Five "gigabit testheds" established by NSFand ARPA (described on page 9) becameoperational. In addition, a DOD-orientedtestbed founded by ARPA focuses on terrainvisualization applications.

a ARPA established a gigabit testhed in theWashington. DC area in cooperation withmore than six other agencies in the area.

Projected FY 19.94 Accomplishments

a Awards will he made to implement a newNSFNET network architecture (includingnetwork access points (NAPS). a routingarbiter, a very high speed backbone, andregional networks).

a Additional very high speed backbones willlink HPCRCs (High Performance ComputingResearch Centers, described on pages 12-14).

8

a Connectivity for DOE's ESnet will grow to27 sites.

a More universal and faster Internet connec-tions for the research and education commu-nity

a Improved network information services andtools

Proposed FY /995 Activities

a NSFNET Implement new architecture(awards were made in FY 1994): implementvery high speed backbone to NSFSupercomputer Centers: establish some addi-tional high speed links for demanding appli-cations

a DOE Upgrade ESnet services to T3 andselected sites to 155 Mb/s: upgrade connec-tivity to Germany and Italy to TI and toRussia to 128 Kb/s (kilobits per second orthousands of bits per second)

a NASA's AEROnet and NSI Establishinternal T3 and higher speed network back-bone to five NASA centers

a Expand connectivity to schools (K-I2through university) connectivity fundedby NSF and NIH will reach a total of 1,500schools. 50 libraries, and 30 medical centers:NASA's Spacelink computer informationsystem for educators will he made availablevia the Internet: toll-free dial-up access willhe provided to teachers without Internetaccess.

a NIH Acquire gigabit (billions of hits persecond) local networks for use with multipleparallel computers and as a backbone toenable development, of the "Xconf" imageconferencing system

a Integrate NOAA's more than 30 environ-mental data centers into the Internet ihniughhigh speed connectivity and new data man-agement tools

=11I.

-I Expand EPA connectixity to reach a substan-tial percentage of Federal. state, and industri-al environmental problem-solving groupsand test distributed computing approaches tocomplex cross-media environmental model-ing

-I Continue to support and improve informa-tion services such as the NSFNET InternetNetwork Information Center (InterNIC)

1.3. Gigabit Speed Networking R&D

New technologies are needed for the new breedof applications that require high performancecomputers and that are demanded by usersacross the U.S. These technologies must movemore information faster in a shared environmentof perhaps tens of thousands of networks withmillions of users. Huge files of data, images.and videos must he broken into small pieces andmoved to their destinations without error, ontime, and in order. These technologies mustmanage a user's interaction with applications.For example. a researcher needs to continuouslydisplay on a local workstation output from asimulation model running on a remote high per-formance system in order to use that informationto modify simulation parameters.

As these gigabit speed networks are deployed.the current harriers to more widespread use of

40-41 AW10ItA11--11 ILIAC*6.4 CASA111-.-111 MAGIC

.11.. KCIAA4A-0 WANE?

HPCC-supported gigabit testheds funded jointlyby NSF and ARPA test high speed networkingtechnologies and their application in the realworld.

9

high performance computers will he surmount-ed. At the same time, high speed workstationsand small and mid-size scalable parallel systemswill gain wider use.

A tercdlops (a trillion floating point opera-tions per second) computing technology baseneeds gigabit speed networking technologies.

The HPCC Program is developing a suite ofcomplementary. networking technologies to takefullest advantage of this increased computationalpower. R&D focuses on increasing the speed ofthe underlying network technologies as well asdeveloping innovative ways of delivering hits toend users and systems. These include satellite,broadcast, optical, and affordable local areadesigns.

The HPCC Program's gigabit testheds areputting these technologies to the test in resource-demanding applications in the real world. Thesetestheds provide working models of the emerg-ing commercial communications infrastructure.and accelerate the development and deploymentof gigabit speed networks.

In FY 1994-1995. HPCC-funded research isaddressing the following:

ATM/SONET (Asynchronous TransferMode/Synchronous Optical Network) tech-nology "fast packet switched" cell relaytechnology (in which small packets of fixedsize can he rapidly routed over the network)that may scale to gigabit speeds

J Interfacing ATM to HiPPI (HighPerformance Parallel Interface) and HiPPIswitches and cross connects to make hetero-geneous distributed high performance com-puting systems available at high networkspeeds

All-optical networking

tt:

High speed LANs (Local Area Networks)

Packetized video and voice and collaborativeworkspaces (such as virtual reality applica-tions that use remote instruments)

a Telecommuting

a Intelligent user interfaces to access the net-work

a Network management (for example. reserv-ing network resources, routing informationover the networks, and addressing informa-tion not only to fixed geographical locationsbut also to people wherever they may he)

Network performance measurement technol-ogy (to identify bottlenecks, for example)

a Networking standards (such as for interoper-ability) and protocols (including networksthat handle multiple protocols such asTCP /IP, GOSIP/OSI, and popular propri-etary protocols)

Additional FY 1995 plans include completingDOE's high speed LAN pilot projects and pro-viding select levels of production-qualityvideo/voice teleconferencing capability. NASAand ARPA plan experiments in mitigating trans-mission delay to the ACTS satellite. NASAplans to extend terrestrial ATM networks toremote locations via satellite and to demonstratedistributed airframe/propulsion simulation viasatellite.

1.4. Network Security

Network data security is vital to HPCC agenciesand to many other users such as the medical andfinancial communities. FY 1994-19'-;.7 researchis directed at incorporating security in the man-agement of current and future networks by pro-tecting network trunks and individual systems.Examples include:

-1Joint ARPA/NSA projects in gigabit encryp-tion systems for use with ATM

10

Use of the ARPA-developed KERBEROSauthentication system by DOE for distributedenvironment authentication and secure infor-mation search and retrieval

a Methods for certifying and accrediting infor-mation sent over the network

NSA is addressing the compatibility of DOD pri-vate networks with commercial public networks.

The rapid growth of networks and of the numberof computers connected to those networks hasprompted the establishment of incident responseteams that monitor and react to unauthorizedactivities, potential network intrusions, andpotential system vulnerabilities. Each teamserves a specific constituency such as an organi-zation or a network. One of the first such teamswas CERT. the Computer Emergency ResponseTeam. based at the Software EngineeringInstitute in Pittsburgh. PA. CERT was estab-lished in 1989 by ARPA in response to increas-ing Internet security concerns, and serves as aresponse team for much of the Internet. FIRST.the Forum of Incident Response and SecurityTeams, was formed under DOD. DOE. NASA.NIST, and CERT leadership. FIRST is a grow-ing global coalition of response teams that alerteach other about actual or potential securityproblems, coordinate responses to such prob-lems. and share information and develop tools inorder to improve the overall level of networksecurity.

2. High Performance Computing Systems

At the beginning of the HPCC Program in 1991,few computer hardware vendors were develop-ing scalable parallel computing systems, eventhough they acknowledged that traditional vectorcomputers were approaching their physical lim-its. By 1993. all major U.S. vendors had adopt-ed scalable parallel technology. Today, a widerange of new computing technologies is beingintroduced into commercial systems that are nowbeing deployed at the HPC'RCs. in industry, andin academia. These include the whole range ofscalable parallel and traditional systems such asfine- and coarse-grained parallel architectures.

...,amm.,

vector and vector/parallel systems. networkedworkstations with high speed interfaces andswitches, and heterogeneous platforms connect-ed by high speed networks. Some of these sys-tems now scale to hundreds of gigaflops (billionsof floating point operations per second). TheHPCC Program is well on track toward meetingits FY 1996 goal of demonstrating the feasibilityof affordable multipurpose systems scalable toteraflops (trillions of floating point operationsper second) speeds.

The architectures of scalable systems how theprocessors connect to each other and to memory.and how the memory is configured (shared ordistributed) vary widely. How these architec-tures communicate with storage systems such asdisks or mass storage and how they networkwith other systems also differ.

In past years. the HPCC Program concentratedon the design and manufacture of high perfor-mance systems, including fundamental underly-ing components, packaging, design tools, simu-lations, advanced prototype systems. and scala-hility. ARPA is the primary HPCC agencyinvolved in developing this underlying scalablesystems technology, often cost-shared with ven-dars, for the high performance computing sys-tems placed at HPCC centers across the country.Efforts are still devoted to developing the foun-dation for the next generation of high perfor-

Simulation of the behavior of materials at the fun-damental atomic scale (adsorption and diffusionof germanium on a reconstructed Si(100 ) sur-face). Simulated using the iPSC/860 hypercubeand the Paragon XP/S-5 supercomputers.

11

mance systems, including new system compo-nents that overcome speed and power limita-tions. scalable techniques to exploit mass storagesystems, sophisticated design technology, andways to evaluate system performance.Additional effort is now being devote(' to devel-oping systems software, compilers, and environ-ments to enable a wide range of applications.

q.

Scaling of memory chip technology is essentialto increase both speed and capacity of comput-er-based systems. This figure shows details of16 memory cells in a high density 1 megabitStatic Random Access Memory (SRAM) thatwere created and visualized using advancedmodel tools for integrated circuit and technologydesign developed at Stanford University. Eachcolor represents a physical layer of material(grey-silicon, yellow-silicon oxide, pink-polysili-con, teal-local interconnect and blue-metal)which has been patterned using advancedlithography and etching techniques. The detailsof geometries and spacings between the variouslayers are critical in determining both the perfor-mance of the SRAM and its manufacturability.Solid geometry models and three-dimensionalsimulations of both materials interactions andelectrical performance are invaluable in optimiz-ing such high density chips.

(Figure Courtesy of Cypress Semiconductor)

ISO

The applications now running on the new sys-tems handle substantially more data bothinput and output than on traditional systems.Graphical display is critical to analyzing thesedata quickly and effectively. For example, out-put from three-dimensional weather models musthe displayed and overlaid with real-time datacollected from networked instruments at remoteobservation stations. Hardware to handle thistask well, such as workstations for scientificvisualization, is part of a high pe'ormance com-puting environment.

Comparison of Josephson Junction technologywith gallium arsenide and CMOS (complemen-tary metal oxide semiconductor) technologiesshowing potential for dramatic improvements inperformance with low power.

The HPCC Program develops and evaluates avariety of innovative technologies that havepotential for future use beyond the next genera-tion of systems. Included in these are supercon-ductive devices. These devices have demon-strated blinding speed and exceptionally lowpower consumption at the chip level, but need tobe scaled up to more complex components to beuseful. If transferable to the system level, thesedevices would have major impact on computingand communications switching systems. NSA isdeveloping a technology demonstration of amultigigabit per second 128x128 crossbar switchthat is potentially expandable to 1000x1000 atvery low power. If successful, the technologywill be evaluated in a system level computingapplication.

12

3. High Performance Computing ResearchCenters (HPCRCs)

HPCRCs are a cornerstone of the HPCCProgram. HPCRCs are the home of a variety ofsystems: early prototypes. small versions ofmature systems, full-scale production systems,and advanced visualiiation systems. The currentproduction systems, capable of hundreds ofgigaflops of sustained throughput, will be suc-ceeded by teraflops systems. These systems arebeing used on Grand Challenge class applica-tions that cannot be scaled down without modi-fying the problem being addressed or requiringunacceptably long execution times. The largestof these applications are being run on multiplehigh performance computers located around thecountry and networked in the gigabit testbeds.

An interdisciplinary group of experts meets atthese centers to address common problems.These include staff from the HPCRCs them-selves, hardware and software vendors, GrandChallenge applications researchers, industrialaffiliates that want to develop industry-specificsoftware, and academic researchers interested inadvancing the frontier of high performance com-puting. Funding is heavily leveraged, withHPCC agencies often contributing discretionaryfunds. hardware vendors providing equipmentand personnel, and affiliate industries payingtheir fair share. Industrial affiliation others a lowrisk environment for exploring and ultimatelyexploiting HPCC technology. Two of theseindustrial affiliations arc:

-I The Oil Reservoir Modeling GrandChallenge in which more than 20 companiesand several universities participate

The High Performance Storage SystemProject in which more than 12 companiesand national laboratories participate

Production-quality operating systems and soft-ware tools are developed at these centers, there-by removing harriers to officient hardware use.Applications software tailored to high perfor-mance systems is developed by early users.many of whom access these systems over theInternet. and increasingly over the gigabit

4 4

--1011=

testbeds. from their workstations. Production-quality applications software is often first run on a Center for Research on ParallelHPCRC hardware. The wide range of hardware Comp,.ation, Rice University, Houston, TXat HPCRCs makes them ideal sites for develop-ing the conventions and standards that enable NASA Centersand test interoperability, and for henchmarkingsystems and applications software. a Ames Research Center. Mountain View. CA

Production-quality applications softwareoften is run first on computing systems atHPCRCs.

The major HPCRCs are

NSF Supercomputer Centers

a Cornell Theory Center. Ithaca. NY

J National Center for SupercomputerApplications. Champaign-Urbana. IL

Pittsburgh Supercomputer Center,Pittsburgh. PA

San Diego Supercomputer Center. SanDiego. CA

Tens of thousands of users from more than 800institutions in 49 states and 1 1 1 industrial part-ners have computed on systems at the NSF cen-ters. Currently there are 8.000 users and 78 part-ners. The centers are developing a NationalMetacenter Environment in which a user willview multiple centers as one. The NationalCenter for Atmospheric Research ( NCAR ) in

Boulder. CO. also receives HPCC funds.

NSF Science and Technology Centers

Center in Computer Graphics and ScientificVisualization Brown University.Providence. RI: CalTech, Pasadena, CA:Cornell University. Ithaca. NY: University ofNorth Carolina. Chapel Hill. NC: Universityof Utah, Salt Lake City. UT

13

w.

Godda.d Space Flight Center. Greenbelt.MD

DOE Centers

Los Alamos National Laboratory. LosAlamos, NM

a National Energy Research SupercomputerCenter. Lawrence Livermore NationalLaboratory. Livermore. CA

Oak Ridge National Laboratory. Oak Ridge.TN

The DOE centers accommodate more than 4,000users from national laboratories, industry, andacademia.

Major systems at HPCRCs include one or moreof each of the following (the number of proces-sors in the largest machine at an HPCRC isshown in parentheses):

a Convex

C3880 (8 vector processors )

Cray Research

C90 (16 vector processors)

T3D (512 processors)

YMP (8 vector processors)

Digital Equipment Corp.

Workstation Cluster

a Hewlett-Packard

H-P Workstation Cluster

IBM

ES9000/900 (6 vector processors)

PVS

SP I (512 processors)

Workstation Cluster

Intel

iPSC 860 (64 processors)

Paragon (5 I 2 processors)

a Kendall Square Research

KSR 1 (160 processors)

a MasPar

MasPar 2 (16,000 processors)

MasPar MP-1 (16,000 processors)

a nCuhe

nCUBE2

a Thinking Machines

CM2 (32.000 processors)

CMS ( 1.024 processors)

Smaller versions of some of these scalable highperformance systems have been installed at morethan a dozen universities. The HPCRCs also usea variety of scientific workstations, such as thosefrom Silicon Graphics and Sun Microsystems,for numerous tasks.

EY /995 Plans

a NSF will install new scalable parallel hard-

14

ware and hardware upgrades, enhanceMetacenter resources, and establish severalmore regional alliances.

a DOE will install two different 150 giganopmachines at two sites.

a NASA will establish a prototype high perfor-mance computing facility comparable innature but not in performance to the ultimateteratlops facility. It will be configured withadvanced high perfc ance machines. earlysystems or advanced prototypes of importantstorage hierarchy subsystems. and sufficient.advanced visualization facilities to enablesystem scaling experiments. NASA GrandChallenge researchers in Federal laborato-ries, industry. and academia will access theseadvanced systems using the Internet andgigabit speed networks. These researcherswill provide a spectrum of experiments forscalability studies. Prototype systems andsubsystem interfaces and protocol standardswill be established and evaluated, accelerat-ing the understanding of the character offuture teraflops computing systems.

a NOAA will acquire a high performancecomputing system for its Geophysical FluidDynamics Laboratory at Princeton, NJ todevelop new scalable parallel models forimproved weather forecasting and forimproved accuracy and dependability ofglobal change models.

a EPA will acquire a scalable parallel systemto support more complex multipollution andcross-media (combined air and water) impactand control assessments.

4. Software

HPCC Program software development effortswere originally planned to address the GrandChallenges associated with agency missions. Asthe Program has matured, these efforts havebeen expanded to support the needs of industryand improve U.S. competitiveness.

=1.9.11

The range of applications software being devel-oped under the Program will assure that highperformance computing systems can be broadlyuseful to the American economy. Now thesesystems must be made easier to use.Experienced software developers and applica-tions researchers, many at or connected toHPCRCs, are working to meet these needs.

It took a decade to develop a collection of effi-cient and robust software for vector machines,and it is widely believed that it will take at leastthat long for parallel systems. Performance isdramatically improving on these systems. due tonew algorithms, systems, and experienced peo-ple. Continued software work is needed to real-ize their full potential. The user community isgrowing so fast that demand for computer timeon these systems exceeds supply.

4.1. Systems Software and Software Tools

The high performance computing environmentmodel involves workstation interaction with highperformance systems. This approach makes itpossible for users to almost transparently accesshigher performance machines as problem sizegrows and the software on these machinestmatures.

This environment is fundamentally and pro-foundly different from and more complicatedthan traditional computing environments. Muchof the systems software and software tools forparallel computing has been redesigned andrewritten to take advantage of the theoreticalbenefit of parallelism and enhance user produc-tivity:

a Operating systems manage dozens to thou-sands of processors. their memory, and net-worked heterogeneous systems.

a New programming languages allow straight-forward expression of parallel constructs.

a Mechanisms express hit manipulation in aparallel en ironment.

15

a Precompilers automatically optimize andparallelize.

a Compilers generate instructions to distributethe computation across the processors, mem-ory. and networks.

a Debuggers help developers find coding mis-takes.

a Performance monitors and displays assistdevelopers in identifying where optimizationefforts might best be spent. facilitate devel-opment of dynamic resource managementstrategies, and are used to evaluate differentarchitectures (in part to recommend changesin vendor-provided utilities).

a Software manages the parallel computer'sinput from and output to other computers.distributed hierarchical mass storage. anddata collection hardware such as satellites.

a New scientific visualization methods andsoftware display the large amounts of dataused and produced by high performancecomputers.

a Software tools enable public disseminationof advanced software and documentation.

a Production environment tools schedule jobs.multitask (run several jobs simultaneously).implement quotas, and provide "checkpointand restart- and on-line documentation.

Evolving conventions and standards enabledevelopers to transport software to differentarchitectures and make it look the same to users.High Performance Fortran (HPF) is an example.Coordinated by the Center for Research onParallel Computation at Rice University. theHPF Forum is a coalition of government. thehigh performance computer industry, and aca-demic groups that is developing standard exien-sions to Fortran on vector processors and mas-sively parallel SIMD (Single InstructionMultiple Data and MIMD (Multiple InstructionMultiple Data) systems.

The software development process is evolution-ary tool developers debug codes and makethem more efficient while users and their appli-cations place new demands on these tools,resulting in further improvements. refinements,speed-ups, and increased user-friendliness.Accelerating this cycle is one objective of theHPCC Program.

4.2. Scientific Computation Techniques

The development and analysis of fundamentalalgorithms for use in computational modelingrepresented in the Grand Challenges are as criti-cal to realizing peak performance of scalablesystems as are improvements in hardware. Suchresearch includes studies of algorithms applica-ble m a wide range of parallel machines as wellas those that take advantage of the strengths ofspecific architectures.

These algorithms address both numerical com-putation (where arithmetic calculations predomi-nate) and non-numeric (where finding and mov-ing data rule). Widely used numerical computa-tions include multidimensional fast Fouriertransforms. fast elliptic and Riemann solvers inpartial differential equations. and numerical lin-ear algebra. The latter includes manipulatingvectors and matrices, solving systems of linearequations. an: computing eigenvalues andeigenvectors. Efficient algorithms attuned tospecialized matrix structure (for example. denseor sparse) are especially sought. Numerical lin-ear algebra is an area that was assumed to hewell understood, having been subject to substan-tial research for vector processors. Somewhatsurprisingly, algorithmic breakthroughs made asthese codes were ported to parallel systems havealso resulted in improved performance on vectorprocessors.

These computations are common to so manyapplications that they are developed by expertsto attain maximal efficiency. and their imple-mentations arc included in general-purposereusable software libraries. When these librariesare updated with the more efficient software.users immediately observe faster execution timesfor their applications. Several HPCC agencies

16

are building such libraries and making themwidely available.

4.3. Grand Challenge Applications

The successful use of scalable parallel systemsfor Grand Challenge applications requiresdesigning new hardware, developing new sys-tems software and software tools, and integratingthese with the idealized setting of mathematicsand with the complex environment of real worldapplications and real world users. The maturingHPCC Program is placing increased emphasis onfacilitating this integration. For example, theProgram sponsored the first "Workshop andConference on Grand Challenge Applicationsand Software Technology" in May 1993. Some250 people representing 34 Grand Challengeteams evaluate.; progress and planned futureactivities. A second workshop is scheduled for1995.

As this new software becomes more efficient,stable, and robust, applications researchers areporting their software to the new parallel sys-tems more quickly and are achieving faster runtimes. They are also obtaining more realisticresults by taking advantage of the faster speeds.larger memory, and the opportunity to add com-plexity, which was not possible before the newarchitectures became available. This realismcomes through:

J Higher resolution (for example. modeling aheat of the human heart at time steps of atenth of a second instead of each second, andat every hundredth centimeter rather thanevery centimeter)

J Faster execution times (for example. modelsthat took days of execution time now takehours, enabling researchers to explore awider range of parameters and time scales

100 year climate models can now he exe-cuted in the same time it used to take for 10year models)

-.1More realistic physics for example.including in weather models the physics 1. atbetter model the effects of clouds

41=

J More realistic models t for example. one"multidisciplinary" model combining sepa-rate atmosphere and ocean models. or onecombining "single discipline" air and waterpollution models)

Entirely new approaches are being developed forcases in which existing models or their algorith-mic components are inappropriate for parallelarchitectures.

Researchers are now henchmarking GrandChallenge applications on a variety of high per-formance systems to determine not only wherethey run fastest but also the reasons why, sincedemand for processors, memory. storage, andcommunication all affect speed. Theseresearchers work closely with systems softwareand software tool developers, and they commu-nicate intensively with hardware vendors. Thisfeedback loop results in a wide range ofimprovements in high performance software andhardware.

As Grand Challenge applications prove the met-tle of scalable parallel architectures, commercialsoftware vendors are becoming more active inmoving their software to these new machines.The success of commercial applications softwareis crucial to the success of the high performancecomputing industry.

NSF, with assistance from ARPA. has funded 16Grand Challenge Applications Groups for threeyears beginning in FY 1993 or FY 1994. DOEhas funded nine multi-year Grand Challengeprojects. some jointly with other 1)01: programs.HPCC agencies. and industry. NASA. N11-1.N1ST. NOAA. and EPA have similar GrandChallenge groups. These groups are addressingproblems in the following areas.

Aircraft

Improved and more realistic models and com-puter simulations of aerospace \ chides andpropulsion systems are being developed. These

17

Simulation of a tiltrotor aircraft (V-22 Osprey)during takeoff. Shown are streaklines, renderedas smoke and computed using UFAT (UnsteadyFlow Analysis Toolkit), a new time-dependentparticle tracing code.

will allow for analysis and optimization ofdesigns over a broad range of vehicle classes,speeds, and physical phenomena, using afford-able, flexible, and fast computing resources.Development of parallel benchmarks is an areaof intensive activity. These applications are com-putationally unrealistic with traditional comput-ing technology. NASA's ComputationalAcroscience Grand Challenges include the HighSpeed Civil Transport, Advanced Subsonic CivilTransport. the High-Performance Aircraft. andRotorcraft. This clearly is an area of significantmutual benefit to both the HPCC Program andother major NASA programs.

In addition, NSF has funded a Grand ChallengeApplications Group to address fundamentalproblems in coupled field problems and geo-physical and astrophysical fluid dynamics turbu-lence.

Computer Science

NSF has funded Grand Challenge ApplicationsGroups in:

a High performance computing for learning

a Parallel I/0 tinput/output) methods for I /O-intensive Grand Challenge applications

.....s411111

VER-.

Energy

DOE's Grand Challenge projects are exploring:

Mathematical combustion modelingdeveloping adaptive parallel algorithms forcomputational fluid dynamics and applyingthem to combustion models

-I Quantum chromodynamics calculationsdeveloping lattice gauge algorithms on mas-sively parallel machines for high energyphysics and particle physics applications

JOH reservoir modeling constructing effi-cient parallel algorithms to simulate fluidflow through permeable media

la) Reptant* (he lopetakm pbme; t. 0.560 atN.

tht Inlet jet Morro °Alston with mapiatiom plate; I. 1.64 ms.

tct Vale. JO= t!ylle.;errodon ot the edge of

td Peek of Wrcttoo tom:1.1.07 IM

Model showing initiation of water heater combus-tion.

18

The numerical Tokamak project develop-ing and integrating particle and fluid plasmamodels on massively parallel machines aspart of the multidisciplinary study ofTokamak fusion reactors

Environmental Monitoring and Prediction

The environmental Grand Challenges includeweather forecasting. predicting global climatechange. and assessing the impacts of pollutants.High performance computers allow better mod-eling of the Earth and its atmosphere. resultingin improved guidance for weather forecasts andwarnings. and improved global change models.

High resolution local and regional weather mod-els are being incorporated into larger nationaland global weather forecasting counterparts.Several of these models are used widely byresearchers to investigate and monitor the behav-ior of the atmosphere through numerical simula-tion. NOAA scientists are redesigning some ofthese models to take full advantage of new scal-:Ible systems. Some global models are "commu-nity models." used by researchers worldwide tocompare results with observations and with otherrelated models, and to evaluate performance.For example. a set of modular, portable bench-mark codes is being developed and evaluated onseveral scalable systems and networked worksta-tions at the Boulder Front Range Consortium.Funding for this work comes from ARPA'sNational Consortium for High PerformanceComputing, NOAA"s Forecast SystemsLaboratory. the NSF-funded National Center forAtmospheric Research, and the University ofColorado. Users of these improved modelsinclude the Federal Aviation Administration andthe National Weather Service.

Models of the atmosphere and the oceans arebeing rewritten in modular form and transportedto several large parallel systems: funding sourcesinclude DOE's Los Alamos NationalLaboratory, NOAAs Geophysical FluidDynamics Laboratory (GFDI ). and the Navy. Amuch greater level of detail now possiblelocal phenomena such as eddies in the Gulf of

amme...saMexico are being modeled, which allows forbetter warning of weather emergencies andimproved design of equipment such as oil rigs.

Separate air and water quality models are beingcombined into a single model which is beingtransported to a variety of massively parallel sys-tems where benchmarks are being used to evalu-ate performance. These models are needed toassess the impact of pollutant contributions frommultimedia sources and to ensure adequate accu-racy and responsiveness to complex environ-mental issues. Nutrient loading in theChesapeake Bay and PCB transport in the GreatLakes are being targeted heginninrz in FY 1994.

In FY 1995. complexities such as aerosols, visi-bility, and particulates will he added: entire envi-ronmental models will he integrated into parallelcomputing em ironments with a focus on emis-sions modeling systems and integration withGeographical Information Systems.

EPA will acquire a large massively parallel sys-tem to he used initially for this research. A tran-sition to an operational computing resource thatsupports improved environmental assessmenttools is planned.

NSF has funded Grand Challenge ApplicationsGroups to address fundamental problems in:

Large scale environmental modeling

The track of Hurricane Emily predicted byNOAA's GFDL (yellow) and the observed track(orange).

Model prediction of the amount of lead, a toxicpollutant, deposited by atmospheric processesduring August 1989.

Adaptive coordination of predictive modelswith experimental observations

JEarthquake ground motion modeling in largebasins

High performance computing for land coverdynamics

Massively parallel simulation of large scale.high resolution ecosystem models

DOE's Grand Challenge projects are exploring:

J Computational chemistry parallelizecodes and develop modeling systems for crit-ical environmental problems and remediationmethods

a Global climate modeling numerical stud-ies with large atmosphere and ocean generalcirculation models

Groundwater transport and remediationdevelop comprehensive multiphase. multi-component groundw ater flow and transportsoftware

NASA has funded Grand Challenge researchteams in:

11? ..et17:1L

J Atmosphere/ocean dynamics and tracechemistr

J Climate models

J Four-dimensional data assimilation for mas-..ie Earth sy stem data analysis

J Discovering 'knowledge in geophysicaldatabases

The envimnmental monitoring and predictionGrand Challenge enables better decisionmaking In. government and industry on issuesthat affect both the economy and the environ-ment.

Molecular Biology and Biomedical Imaging

FY 1994 NIH accomplishments include:

J Development and field testing of theDiagnostic X-ray Prototype Network(DXPnet), a nationwide radiology researchimage system

Deployment of Network Entrez. an Internet-based client-server system that enables inte-grated searching of DNA and proteinsequences and the medical literature linkedto those sequences

-I Usage of the Internet-based BLAST (BasicLocal Alignment Sequence Tool) forath anced biological similarity searchingreached million queries per year.

-31mproed understanding of heart functionthrimgh the ihree-dimensional simulation ofa single heartbeat (this required 150 hours onthe fastest Cray received a SmithsonianAward)

i First simulation of an entire biological mem-

20

A closeup view look:.., m on the aortic valvein a computational model vf olocd flow in theheart. The model was developed by CharlesPeskin and David McQueen of New YorkUniversity and run on the PittsburghSupercomputing Center's Cray C90.

brane including all lipid and protein compo-nents, enabling better understanding of themechanism of inflammation of tissues in dis-eases such as asthma and arthritis (collabora-tive work with Eli Lilly)

J Order of magnitude speedups in severalmolecular analysis algorithms

J New software for coupling vector processorswith massively parallel systems, providing anew avenue for molecular dynamics calcula-tions

New algorithms for registration and render-ing of three-dimensional images from two-dimensional clinical images and micrographs

J Assistance in the design of new drugs toinhibit HIV replication

I Further elucidation of the structure of theherpes virus using high performance comput-ing to obtain three-dimensional images fromelectron micrographs

I Determined the three-dimensional structureof several proteins by using a genetic algo-rithm approach to automate the spectralassignment task in NMR spectroscopy

-I Developed parallel mo:,.3_ r dynamics sim-ulacon software to detern. the effects ofhydration on protein structure

-I Developed benchmark codes for evaluatinghigh performance systems

FY 1995 NIH plans include:

-I Expand research in algorithms for real-timeacquisition and processing of multi-modelmedical images for use in tkiemedicine andvirtual environments for surgery

Develop new algorithms and software forreceptor-based drug design: study basicmechanisms of receptor structure and func-tion: and develop software to model specificpopulations at risk for disease

R&D in medical image processing for basicresearch, clinical research, and health caredelivery

Support development of telemammography,addressing high resolution, wide field-of-view displays and high performance. lowcost networks for image transmission

NSF has funded Grand Challenge ApplicationsGroups in:

Biomolecular design

J Imaging in biological research

I Advanced computational approaches tobiomolecular modeling and structure deter-mination

Understanding human joint mechanismsthrough advanced computational models

DOE's Grand Challenge projects are exploringcomputational structural biology understanding the components of genomes and developing

21

a parallel programming environment for struc-tural biology.

Product Design and Process Optimization

Beginning in FY 1995. NIST will develop newhigh performance computing tools for applyingcomputational chemistry and physics to thedesign, simulation, and optimization of efficientand environmentally sound products and manu-facturing processes. Initial focus will be on thechemical process industry, microelectronicsindustry, and biotechnology manufacturing.Existing software for molecular modeling andvisualization will be adapted for distributed.scalable computer architectures. New computa-tional methods for first-principles modeling andanalysis of molecular and electronic structure.properties, interactions , nd dynamics will hedeveloped. Macromolecular structure, molecu-lar recognition. nanoscale materials formation,reacting flows, the automated analysis of com-plex chemical systems. and electronic and mag-netic properties of thin films will receive particu-lar emphasis.

Diamond-tool turning and grinding machines arethe epitome of precision manufacturing tools,capable of machining high precision optical finish-es without additional polishing (such as on thiscopper mirror for a laser system). NISTresearchers and their industrial partners aredeveloping methods for monitoring and control-ling diamond turning machines to improve theprecision and production of highly efficient opticssuch as mirrors for laser welders.

NSF has funded a Grand Challenge ApplicationsGroup to address fundamental problems in highcapacity atomic-level simulations for the designof materials.

A DOE Grand Challenge project involves first-principles simulation of materials propertiesusing a hierarchy of increasingly more accuratetechniques that exploit the power of massivelyparallel computing systems.

Space Science

NASA's Earth and Space Science projects sup-port research in:

-I Large scale structure and galaxy formation

-1Cosmology and acc-!tion astrophysics

Convective turbulence and mixing in astro-physics

-I Solar activity and heliospheric dynamics

NSF has funded Grand Challenge ApplicationsGroups to address fundamental problems in:

-I Black hole binaries: coalescence and gravi-tational radiation

-I The formation of galaxies and large-scalestructure

-I Radio synthesis imaging

5. Technologies for the NII

The HPCC Program is helping to develop muchof the technology underlying the Nil in order toaddress National Challenge problems of signifi-cant social and economic impact. This technolo-gy includes advanced information services, soft-ware development environments, and user inter-faces:

22

5.1. Information Infrastructure Services

These services provide the underlying buildingblocks upon which the National Challenges canhe constructed. They provide layers of increas-ing intelligence and sophistication on top of the"communications bitways." These include:

-1 Universal network services. These areextensions to existing Internet technologythat enable more widespread use by a muchlarger user population. They include tech-niques for improved ease-of-use. "plug andplay" network interoperation, remote mainte-nance, exploitation of new "last mile" tech-nologies such as cable TV and wireless,management of hybrid/asymmetric networkbandwidth, and guaranteed quality of servicefor continuous media streams such as video.

-I Integration and translation services. Thesesupport the migration of existing data files,databases, libraries, and software to new,better-integrated models of computing suchas object-oriented systems. They providemechanisms to support continued access toolder "legacy" forms of data as the modelsevolve. Included are services for data formattranslation and interchange as well as tools totranslate the access portions of existing soft-ware. Techniques include "wrappers" thatsurround existing elements with new inter-faces, integration frameworks that defineapplication-specific common interfaces anddata formats. and "mediators" that extendgeneric translation capabilities with domainknowledge-based computations. permittingabstraction and fusion of data.

-1 System software services. These includeoperating system services to support com-plex. distributed, time-sensitive, and band-width-sensitive applications such as theNational Challenges. 'They support the dis-tribution of processing across processingnodes within the network: the partitioning ofthe application logic among heterogeneousnodes based on their specialised capabilitiesor considerations of asymmetric or limitedinterconnection bandwidth: guaranteed real-

%Mt

:;:-'4=411:111M111111111111MINMINIMINV

time response to applications for continuousmedia streams; and storage, retrieval, andI/O capabilities suitable for delivering largevolumes of data to very large numbers ofusers. Techniques include persistent storage.programming language support, and file sys-tems.

-I Data and knowledge management services.These include extensions to existing databasemanagement technology for combiningknowledge and expertise with data. Theseinclude methods for tracking the ways inwhich information has been transformed.Techniques include distributed databases,mechanisms for search, discovery, dissemi-nation, and interchange, aggregating basedata and programmed methods into"objects." and support for persistent objectstores incorporating data, rules, multimedia,and computation.

Information security services. These help inprotecting the security of information.enhancing privacy and confidentiality, pro-tecting intellectual property rights, andauthenticating information sources.Techniques include privacy-enhanced mail.methods of encryption and key-escrow. anddigital signatures. Also included are tech-niques for protecting the infrastructure (suchas authorization mechanisms and firewalls)against intrusion attacks (such as by worms.viruses, and trojan horses).

Reliable computing and communicationsservices. These include services for non-stop. highly reliable computer and communi-cations systems operating 24 hours a day. 7days a week. The techniques include mecha-nisms for fast system restart such as processshadowing. reliable distributed transactioncommit protocols. and event and data redologging to keep data consistent and up-to-date in the face of system failures.

5.2. Systems DevekpmentEnvironments

These will provide the network

and Support

-based software

23

development tools and environments needed tobuild the advanced user interfaces and the infor-mation-intensive National Challenges them-selves. These include:

-1 Rapid system prototyping. These consist ofsoftware tools and methods that enable theincremental integration and cost effectiveevolution of software systems. Technologiesinclude tools and languages that facilitateend-user specification. architecture designand analysis, and component reuse and pro-totyping; testing and on-line configurationmanagement tools; and tools to support theintegration and interoperation of heteroge-neous software systems.

-I Distributed simulation and synthetic envi-ronments. These sf.-,ftware developmentenvironments support the creation of synthet-ic worlds that can integrate real as well asvirtual objects that have both visual andcomputational aspects. Methods includegeometric models and data structures; toolsfor scene creation, description, and anima-tion; integration of geometric and computa-tional models of behavior into a combinedsystem description; and distributed simula-tion algorithms.

-1 Problem solving and system design environ-ments. These provide automated tools forsoftware and system design that are flexibleand can be tailored to individual needs.Examples include efficient algorithms forsearching huge planning spaces; more pow-erful and expressive representations of goals.plans. operators, and constraints; and effi-cient scheduling and resource allocationmethods. The effects of uncertainty and theinteractions of goals will he addressed.

-1 Software libraries and composition support.Common architectures and interfaces willincrease the likelihood of software reuseacross different computational models, pro-gramming languages. and quality assurance.By developing underlying methodology. datastructure, data distribution concepts. operat-ing systems interfaces, synchronization fea-

JO

=rIi..00.111.0.11NINIMIIIKS

.1111101Ol.....111011

tures. and language extensions. scalablelibrary frameworks can he constructed.

-I Collaboration and group software. Thesetools support group cooperative work envi-ronments that span time and space. Theywill make it possible to join conferences inprogress and automatically he brought up todate by agents with memory. Methodsinclude network-based video conferencingsupport, shared writing surfaces and "livehoards." document exchange, electronic mul-timedia design notebooks, capturing designhistory and rationale, agents or intermedi-aries to multimedia repositories, and versionand configuration management.

5.3. Intelligent Interfaces

Many of the National Challenge applicationsrequire complex interfacing between humansand intelligent control stems and sensors, andamong multiple control systems and sersors.These applications must understand their envi-ronment and react to it. High level user inter-faces are needed to satisfy the many differentrequirements and preferences of vast numbers ofcitizens who will interact with the Nil.

-I Human-computer interface. A broad rangeof integrated technologies will allow humansand computers to interact effectively, effi-ciently. and naturally. Technologies will hedeveloped for speech recognition and gener-ation: graphical user interfaces will allowrapid browsing of large quantities of data:user-sensitive interfaces will customize andpresent information for particular levels ofunderstanding: people will use touch, facialexpressions, and gestures to interact withmachines: and these technologies will adaptto different human senses and abilities.These new integrated. real-time communica-tion modalities will he demonstrated in mul-timedia. multi-sensory environments.

Heterogeneous database interfaces.Methods to integrate and access heteroge-

24

neously structured databases composed ofmulti-formatted data will he developed. In afuture Nil's information dissemination envi-ronment. a user could issue a query that isbroadcast to appropriate databases and wouldreceive a timely response translated into thecontext of the query. Examples of multi-for-matted data include ASCII text, data that areunivariate (such as a one-dimensional timeseries) or multivariate (such as multi-dimen-sional measurement data), and time series ofdigital images (such as a video).

Image processing and computer vision.Images. graphics, and other visual informa-tion will become more useful means ofhuman-computer communication. Researchwill address the theory, models, algorithms.architectures, and experimental systems forlow level image processing through highlevel computer vision. Advances in patternrecognition will allow automated extractionof information from large databases such asdigital image databases. Emphasis is placedon easily accessing and using visual informa-tion in real-world problems in an environ-ment that is integrated and scalable.

User-centered design tools/systems. Newmodels and methods that lead to interactivetools and softwi..0 systems for user-centeredactivities such as design will he developed.Ubiquitous. easy-to-use. and highly effectiveinteractive tools are emphasized. A newresearch area is user-friendly tools that com-bine data-driven and knowledge-based capa-bilities.

-I Virtual !.eality and telepresence. Tools andmethods for creating synthetic (virtual) envi-ronments to allow real-time. interactivehuman participation in the computing/com-munication loop will be addressed.Participation can he through sensors, effec-tors, and other computational resources. In

support of National Challenge applicationareas. efforts will focus on creating sharedvirtual environments that can he accessedand manipulated by many users at a distance.

;

6. National Challenges

These are large-scale, distributed applications ofhigh social and economic impact that contain anextensive information-processing componentand that can benefit greatly by building anunderlying information infrastructure. NationalChallenges to he addressed by the HPCCProgram in FY 1994 and FY 1995 include:

Digital Libraries

A digital library is the foundation of a knowl-edge center without walls, open 24 hours a day.and accessible over a network. The HPCCProgram supports basic and strategic digitallibraries research and the development anddemonstration of associated technologies. Thesetechnologies arc used in all of the other NationalChallenge applications.

Beginning in FY 1994. the Program will supportthe following R&D. much of it funded by a jointNSF/ARPA/NASA "Research in DigitalLibraries" initiative:

-iTechnologies for automatically capturingdata of all forms (text. images. speech.sound. etc.). generating descriptive informa-tion about such data (including translationinto other languages). and categorizing andorganizing electronic information in a varietyof formats.

Advanced algorithms and intelligent interac-tive Internet-based tools for creating andmanaging distributed multimedia databasesand for browsing. navigating. searching. fil-tering. retrieving. combining. integrating.displaying. visualizing. and analyzing verylarge amounts of information that are inher-ently in different formats. These databasesare frequently stored on different media thatare distributed among heterogeneous systemsacross the Nation and around the world.Research will also address standards thatenable interoperability.

NOAA will provide Internet-based access to and

distribution of remote sensing imagery and othersatellite products from its geostationary.andpolar orbiting operational environmental remotesensing satellites. NASA will provide access toother remote sensing images and data over theInternet and the gigabit testheds. ThiS includesmaking observational data from satellites avail-able to state and local governments, the agricul-ture and transportation industries, and to librariesand educational institutions.

NSA will develop a prototype environment ofthe future in which a user, an application level-oiler, and a data administrator each sees an inte-grated information space in terms directly mean-ingful and accessible to them, rather than as acollection of relatively unintelligible, difficult-to-access databases.

Crisis and Emergency Management

Large-scale, time-critical, resource-limited prob-lems such as managing natural and man-madedisasters are another vital National Challenge.Effective management involves the use of com-mand. control, communications, and intelligenceinformation systems to support decision makersin anticipating threats. formulating plans, andexecuting these plans through coordinatedresponse. Many other National Challenge pro-jects provide information and information man-agement tools for use in crisis and emergencymanagement. HPCC efforts include ARPA'sresearch projects on ubiquitous data communica-tions infrastructure in the face of disasters.including the timely development and transmis-sion of plans to operational units. exploitation oftechnical and human sources of information, andinput to command. NOAA plans to make avail-able environmental warnings and forecasts andother relevant information to support emergencymanagement through the Internet. In FY 1994.NSF initiated a program to support researchleading to development of information infras-tructure technologies that can he integrated intothe civil infra..tructure, including transportation,water quality. safety of waste removal. andaccess to energy sources.

Education and Lifelong Learning

HPCC support for this National Challengeinvolves making HPCC technologies a resourcefor the Nation's education, training, and learningsystems for people of all ages and abilitiesnationwide. The NII approaches this challengefrom several directions:

Distance learning will bring specializedresources in a timely manner to geographi-cally widespread students.

Teacher training and coordination enhancesthe resources available to teachers at all edu-cational levels.

Students throughout the country will haveaccess to information and resources previ-ously only available at research and librarycenters.

Lifelong learning provides educationalopportunities to populations regardless ofage or location.

Digital libraries will make information avail-able throughout the network both for pro-fessionals as well as students at all levels.

The HPCC Piogram is providing network accessand conducting pilot projects that demonstrateHPCC technologies for improvirg learning andtraining and that can he scaled to nationwidecoverage. A program in networking infrastruc-ture for education was begun by NSF in FY1994. On-going K-12 programs in science. engi-neering. and biomedical and health applicationsare conducted by almost all HPCC agencies.

Electronic Commerce (EC)

This National Challenge integrates communica-tions. data management. and security services toallow different organizations to automaticallyexchange business information.Communications services transfer the informa-tion from the originator to the recipient. Datamanagement services define the interchange for-mat of the information. Security services

26

authenticate the source; verify the integrity ofthe information received: prevent disclosure byunauthorized users; and verify that the informa-tion was received by the intended recipient.Electronic commerce applies and integratesthese services to support business and commer-cial applications such as electronic bidding.ordering and payments. and exchange of digitalproduct specifications and design data.

ARPA will develop a common underlyinginfrastructure for authentication, authorization.accounting and banking services, usage meter-ing, and fee-for-access within networks and dis-tributed systems. ARPA is also developingmechanisms for active commerce that will seekout qualified bidders on behalf of customers.based on extensive knowledge of the bidderscapabilities and the customers needs.

Beginning in FY 1995. NIST will collaboratewith industry to develop and apply technologiesthat enable electronic commerce in general, withinitial emphasis on the manufacturing of elec-tronic and mechanical components and subsys-tems. The agency will conduct R&D in securityservices and establish facilities to support inter-operability testing.

Energy Management

Oil consumption. capital investment in powerplants. and foreign trade deficits all benefit fromimproved management of energy demand andsupply. Beginning in FY 1994, DOE and thepower utilities will document and assess thetools and technologies needed to implement theNational Challenge of energy demand and sup-ply management. They will also document theexpected economic benefits and identify policyor regulatory changes needed so that the utilitiescan participate in the deployment of the NII.

Environmental Monitoring and WasteMinimization

Improved methods and information will dramati-cally increase the competitiveness of V.S. com-panies in the world's S I 00 billion per year envi-

13;3

One Giant Leap ... Networks: Where Have You Been All My Life?

Midway through my junior year at New Hanover High School in Wilmington. North Carolina. an experience began for me that has re-routedthe path of my entire education and learning adventures. I. and three other students. won a national scientific computing contest calledSuperQuest, sponsored by the Cornell University and the National Science Foundation. SuperQuest was no ordinary science fair -- rather,it was a "take a giant leap outside of your mind" contest. It was a fortuitous opportunity for us high schoolers. Our sudden introduction tothe world of high performance computers included IBM RISC clusters. an ES/9000 vector processor. and a KSR parallel processor.

One of the greatest benefits of participating in the SuperQuest program has been my exposure to computer networks and telecommunica-tions. I delved into the online world for my first time before we had won, when the team and ! were constructing a science project theSuperQuest judges might deem a winner. We spent just a semester reaching consensus on the topic! Each of us had ideas from inves-tigating mag-lev trains, to orb spiderwebs, to the pitching of a baseball. Which of these were within our capability? And which should weperhaps leave to the Princeton researchers? We turned to the Internet to gather the advice and experience of more knowledgeable folks.Our first action was to post questions on as many science bulletin boards and online services as possible. The two we frequented were theSuperQuest homebase at the Cornell Theory Center. and the High Performance Computing network (HPC Wire) in Colorado. Both werethe ideal sources to check the feasibility of modeling an orb spider web or the flight of a baseball on a supercomputer. Sometimes wereceived answers in a day -- and sometimes within an hour. We were wowed not only by the speed of the knowledge transfer, but also,bythe altruism of the network community. Through our communications, we were able to quickly focus on the orb spider web as our project.It was within our scope.

In the archaic tradition of our 12 years of schooling, we trekked to the local library to research spider webs. This proved to be timecon-suming. One of our teachers introduced us to WAIS, an Internet ;ndexed database search function. Ecstasy! We sat with him as helogged onto the net from his desktop PC, called up WAIS, and ordered it to do both a worldwide search of indexes with the words "spider"or "web" in them, and a cross-reference follow-up. Within 20 minutes. the net had spewed 20 pages of information sources. We were filledwith the the sudden comprehension of the scientific proces -- the gathering of data and the elimination of possibilities.

As part of the SuperQuest prize. we attended a three-week summer seminar at the Cornell Theory Center. There we were introduced toLAN networks and file-sharing, mainframes and X-terminals, and the minute details of the Internet. Thus, we could suddenly access thelibraries of each of the seven schools at Cornell. as well as data from specialized projects such as the synchrotron and recent biomedicalresearch. Again. the process of gathering data was accomplished primarily through using computer communications.

Upon our return home. we were able to use our Internet connection (which was another SuperQuest reward) to continue our research andalso access the supercomputer facilities back at Cornell After two months of a maniacal pace, and with our interim report mailed. wedecided to take a week-long break. However. I found myself drawn to the computer --there were so many interesting things to explore --and I was so curious. I logged into "Sunsite" at the University of North Carolina at Chapel Hill and began browsing archives, just for the funof it. I pulled up a picture of an ancient Vatican manuscript, with its crinkled brown pages and mottled writing. However, disappointed that Icould not make out any of the words. I was about to close the window, when i had an idea! Using the XV software (also obtained from thenet). I zoomed into a word and smoothed it into something legible. I had stumbled into a combination of resources that allowed me toexamine the document.

And I didn't stop there. My school system's current agenda involves the reforming of the traditional daily schedule: we may initiate "block"schedules in the coming year. As student body president. I represent the school on the Superintendent's Advisory Board -- our mission: toinvestigate the pros and cons of block scheduling. We decided that most important, we needed to hear from students in other schools thathad implemented the program. The nearest one was five hours away. A group of teachers had been bused in the previous month to visitfor an hour or two of questions the only time available after travel time. No student on the advisory board wanted to repeat this exerciseand in fact, we were almost reduced to drawing straws for a victim, when I had a sudden flash. A bit of background: The state of NorthCarolina has. within the past five years, set up a fiberoptics network that enables a teacher at one school to simultaneously each her homeclass as well as classes that are located across the county and even the state. My moher was teaching an oceanography class over thesystem and I knew that at least one of those schools was block-scheduled. At the next advisory board meeting. I moved that we postponeour road trip and use the.fiberoptics network for an afterschool telecon-ference. Within four weeks. we had teleconferenced with two block-scheduled schools. We asked questions that mattered to students:what happens when you miss school: how are athletics affected: and.how are advanced placement classes organized. Moreover, the twoblock-scheduled schools were able to discuss their own variations ofblock scheduling

Computer technology has not simply affected my education; it haschanged my personality. I have travelled from having a daydreamabout why spider webs are so strong. to performing concrete scientificresearch. to making an obscure document understandable (one that Idid not know existed. but for the Internet). and I initiated a solution to areal-life organizational problem. I'm feeling pretty good

Frank *Gib' Gibson 1579,1 national !ugh school wumei of the NSA NASA EDtelecommunications essay with teacher. Abigail Saxon

27

44.

lir,/11

--1.&1211811111

ronmental monitoring and waste managementindustries. Beginning in FY 1995, digitallibraries of the large volume and wide range ofenvironmental and waste information will heassembled and tools will he developed to makethese libraries useful. These include:

JDOE site survey and regulatory informationand tools to use these libraries (the DOE'sexisting weapons complex will serve o a

natural testhed)

A coordinated effort by NASA, NOAA. andEPA will jointly provide public access to awide variety of Earth science databases.including satellite images. Earth sciencemeasurements, and in situ and satellite datafrom NOAA's environmental data centers.

-IA joint NASA/NOAA/EPA effort will devel-op training and education to satisfy the pub-lic's environmental information needs.

- I A National Environmental InformationIndex. as directed by the NationalPerformance Review

A NOAA Earth Watch pilot information sys-tem will provide integrated access to envi-ronmental information together with relevanteconomic and statistical data for policy mak-ers and others.

Health Care

Advanced HPCC communication and informa-tion technologies promise to improve the quality,effectiveness and efficiency of today's healthcare system. This National Challenge will com-plement the biomedical Grand Challenges. Itwill include testhed networks and collaborativeapp:ications to link remote and urban patientsand providers to the information they need,database technologies to collect and sharepatient health records in secure, privacy-assuredenvironments, advanced biomedical devices andsensors, and system architectures to build andmaintain the complex health information infras-

28

tructure.

Using a Broad Agency Announcement. NIHbegan funding the following activities in FY1993, and is expanding efforts beginning in FY1994:

Testhed networks to link hospitals, clinics.doctor's offices. medical schools, medicallibraries, and universities to enable healthcare providers and researchers to share medi-cal data and imagery

Software and visualization technology tovisualize the human anatomy and analyzeimages from X-rays. CAT scans, PET scans,and other diagnostic tools

Virtual reality technology to simulate opera-tions and other medical procedures

-1Collabotative technology to allow severalhealth care providers in remote locations toprovide real-time treatment to patients

-I Database technology to provide health careproviders with access to relevant medicalinformation and literature

Database technology to store, access, andtransmit patients' medical records while pro-tecting the accuracy and privacy of thoserecords

A three-year contract was awarded in FY 1993to a consortium of nine West Virginia institu-tions to use advanced networking technologiesto deliver health services in both rural and urbanareas. Other proposals received in response tothe announcement will he funded in FY 1994.

Beginning in FY 1995, NIH will provide cancerprevention and treatment information to the pub-lic via several multimedia systems includingMosaic (described on pages 33-341.

An ARPA Biomedical Program will developadvanced biomedical devices and tools to buildnext-generation health care information systems.

3i;

NSF will expand a program in health care deliv-ery systems begun in FY 1994. Included areactivities in cost-effective telemedicine systemsfor distance medicine applications.

Manufacturing Processes and Products

Advancing manufacturing through the use ofHPCC technologies in design. processing. andproduction of manufactured products is anotherNational Challenge. A key element is the devel-opment of the infrastructure technology andstandards necessary to make the processes andproduct information accessible to both enterpris-es and customers. This Challenge relies on net-work security and on the Digital Libraries andElectronic Commerce National Challenges. andis closely related to the Energy ManagementChallenge. On-going multi-year projectsinclude:

ARPA's MADE (Manufacturing Automatonand Design Engineering) in AmericaProgram to develop engineering tools andinformation integration capabilities to sup-port future engineering and manufacturingprocesses. These include the Center forAdvanced Technology. a joint industry-gov-ernment facility that offers world class man-ufacturing technology training: centers andnetworks for agile manufacturing: and a vir-tual library for engineering.

Beginning in FY 1995. expandedNASA/industry/academia efforts in the mul-tidisciplinary design of aeronautical air-frames and aircraft engines will develop anintegrated product/process developmentcapability. This is intended to shorten theproduct development cycle, maximize capa-bility. lower the life cycle cost, and obtainnew insight into and understanding of theadvanced manufacturing process. all criticalto more competitive airframe and propulsionindustries.

NIST's System Integration forManufacturing Applications Programemphasizes technologies that support flexi-ble and rapid access to information for man-

29

ufacturing applications. It includes a stan-dards-based data exchange effort for com-puter integrated manufacturing that focuseson improving data exchange among design.planning, and production activities. Resultswill he made available to U.S. industrythrough workshops. training materials, elec-tronic data' repositories, and pre-commercialprototype systems.

J At NIST, an Advanced ManufacturingSystem and Network Testhed supports R&Din high performance manufacturing systemsand testing in a manufacturing environment.The testhcd will he extended to include man-ufacturing applications in mechanical, elec-tronics, construction, and chemical indus-tries: electronic commerce applications formechanical and electronic products: and anintegrated Standard Reference Data system.The testbed will serve as a demonstrationsite for use by industrial technology suppli-ers and users, and to assist industry in thedevelopment and implementation of volun-tary standards.

NSF will support research leading to thedevelopment of information infrastructuretechnologies that support manufacturingdesign. Initial focus will he on virtual andrapid prototyping.

Public Access to Government Information

This National Challenge will vastly improvepublic access to information generated byFederal, state, and local governments throughthe application of HPCC technology. On-goingefforts include connecting agency depositorylibraries and other sources of government infor-mation to the Internet to enable public access:and demonstrating. testing. and evaluating tech-nologies to increase such access and effectiveuse of the information.

For example, the White House. the U.S.Congress. and the HPCC Program make infor-mation available on the Internet. Other exam-ples are ARPA's research in delivering comput-er science reports and literature to researchers

...... %MI

'-.. i-1.-.:6',Thilood .,e 4 11" foci'Th;,

' Oa . Me ' ' . 1,-,Fr- '..t'e,:f-

ult. .,,,,,,0,1,,ord,d e-,',"11 .......aiii.z1.r.;:;.:p1=__,*.,..?i,lw-4,

;OA

The concept of the Nll's "InformationSuperhighway" has captured the imagination ofthe Nation.

and the public: NSF's Science and TechnologyInformation Service and its support for theSecurities and Exchange Commission's Edgarsystem: NSF's pilot project to demonstrate theuse of the Internet and Mosaic in disseminatingNSF information about program activities andaccomplishments: DOE efforts to make energystatistics available to the public:NASA/NOAA/EPA efforts to make environ-mental data available to researchers and the gen-eral public: and Public Health Service (PHS)sponsorship of a variety of electronic informa-tion services, including NIH and NLM Internetservers that also provide connectivity to otherhealth information services such as the bulletinboard at the Food and Drug Administration andthe PHS AIDS bulletin hoard.

7. Basic Research

Basic research projects focus on de\ eloping newmethods to address fundamental limitations inHPCC technology as the Program proceeds andensuring that the foundations for the next gener-ation of HPCC technology are developed. Muchof the advanced basic research is carried out inthe academic communit in cooperation withindustr.

J ARPA funds basic research coupled to its()ther elforts in high perforinance computing.Basic research areas include design science.

30

human/computer interaction. human lan-guage technology, persistent object bases.and software foundations. In addition,ARPA funds a high performance computinggraduate fellowship program to focus atten-tion on the critical need for people trained inthis field.

a NSF funds long-term investigator-initiatedresearch and continues to encourage interdis-ciplinary research, collaboration betweencomputer scientists and applications scien-tists in solving Grand and NationalChallenges, and cross-sector partnerships. It

funded 350 investigator-initiated projectsand awarded 77 postdoctoral research andtraining grants (one grantee was a 1993Supercomputing Forefront Award Winner).NSF-supported researchers contribute to fun-damental networking, memory, interconnec-tivity, storage. and compiler technology, andthe agency plans to support rP ,-!arch in virtu-al reality. NSF. ARPA. zinc, NASA jointlyfunded High Performance Fortran develop-: neat.

a In FY 1995 NSF plans to support 100 newinvestigator-initiated projects in new areas.support 30 new postdoctoral fellows. and ini-tiate three new programs graduate fellow-ship (20 awards initially). Industry/HighPerformance Computing Centers visitor pro-gram (16). and Software InfrastructureCapitalization (2).

a NSF also supports the procurement or scal-able parallel systems for has' research andin FY 1995 will support int, aructure forNational Challenges.

a DOE funds basic research at agency labora-tories and at 30 universities including over-40 postdoctoral associates and over 60 grad-uate students. Subjects include numericalanalysis and scientific computing. modelingand analsis of physical systems. d numicalsystems theory and chaos, gcomorie andsyinholic computation, and optimization the -or) and mathematical programming.

NASA sustains research efforts in architec-

tures. algorithms, networked distributedcomputing, numerical analysis. and in appli-cations-specific algorithms. It has researchinstitutes and centers of excellence at theIllinois Computer Laboratory for AerospaceSystems and Software, and at its Ames.Langley. and Goddard centers. It is expand-ing support for postdoctoral research, newprofessors, and its Graduate StudentResearchers Program at NASA centers.

NIH has formal degree-granting fellowshipsin medical informatics and cross-disciplinarytraining of established investigators. Theagency sponsored hands-on training ofbiomedical researchers in using computa-tional biology tools at NSF SupercomputerCenters.

J EPA supports cross training of computation-al and environmental scientists.

8. Training and Education

A natural consequence of basic research inHPCC technology is education and career devel-opment. Each generation of HPCC researcherstrains and educates the next generation. Theworkforce becomes increasingly technologicallysophisticated. providing myriad economic andsocial benefits to the Nation.

J NSF Supercomputer Centers conduct some200 training events for 3,000 trainees eachyear. The agency provided about 20 com-puter systems to colleges and universities.including minority institutions, and fundedfive SuperQuest teams each with four stu-dents and two teachers investigating theirown scientific projects: over the years 64such projects have been funded. In FY 1995the agency will provide 13 more entry-levelsystems to universities and expandSuperQuest.

NSF provides on-going support for trainingand education programs for teachers and stu-dents at all line's. In their program support-ing VLSI (very large scale integration) fabri-

31

Father and daughter using the KidPix graphicsprogram on a Macintosh during "SDSC Kids'Day" at the San Diego Supercomputer Center.

cation by Mosis (metal oxide semiconductorimplementation service). NSF and ARPAfund 1.200 student projects per year: over sixyears 30,000 students from 170 universitiesin 48 states and the District of Columbiahave been funded. The agency supports pilotprojects demonstrating the application ofadvanced technologies in education.Included are the "Common Knowledge" pro-ject involving the Pittsburgh school district.the "National School Network" involvingteachers and students -it all pre-college lev-els, and the "Learning through CollaborativeVisualization" testbed.

i DOE educational programs includeAdventures in Supercomputing for in-serviceteacher training (50 teachers and 25 schooldistricts in five states in FY l' '95). Superkidsfor high school student summer enrichment.and undergraduate and graduate electronictextbook projects.

J NASA supports undergraduate and graduatelevel training. New NASA mechanisms willsupport students and new faculty interestedin applying HPCC technology. especially bydirectly funding students with advisers inter-ested in NASA applications.

NASA w ill continue its K -12 EducationalOutreach Program at se\ en NASA field cen-

1

MiMilliiillini.IMMIIMIIIIIIIMIIIIMIlliMill_TIMM MY illtern nationwide to develop curriculum prod-ucts and teaching aids in computational sci-ence and networking for education.Teachers located near the field centers willparticipate in each project. Finished prod-ucts will he available to all teachers and stu-dents via the Internet.

JARPA has a unique program supporting his-torically black colleges and universities:strongly encourages affiliation with localcommunities: and is accelerating technologytransition through community-wide effortssuch as the National Consortium for HighPerformance Computing.

-INIH will support development of interactivelearning tools for improved science educa-tion. The agency has a pilot project to edu-cate high school science toachers and stu-dents about computational techniques forbiomedical science.

J NOAA provides its agency staff with educa-tion and training in the use of scalable sys-tems.

-1 EPA is developing a prototype training pro-gram that includes a series of pilot projectswith Federal, state, and industrial environ-mental groups. The agency funds fellow-ships and graduate student support in envi-ronmental modeling, supports undergraduatestudents. and supports environmental learn-ing experiences for junior and senior highschool students. It will develop environmen

al education tools for grades 9 through 12.

Through these efforts. more students will choosecareers in HPCC technology and its application,resulting in more widespread and advancedknowledge of these fields. The long term bene-fits of these education programs will includeincreased awareness and education of the gener-al public about high performance computing andcommunications and the application of HPCCtechnology to improve the quality of life in theU.S.

32

9. HPCC Program Management

The HPCC Program reports to the Director ofthe Office of Science and Technology Policy.At his direction, overall Program budget over-sight is provided by the National Science andTechnology Council through its Committee onInformation and Communication. In September1993. the National Coordination Office for HighPerformance Computing and Communications(NCO) completed its first year of coordinatingthe HPCC Program and serving as a liaison tothe U.S. Congress. state and local governments.foreign governments. industry, universities, andthe public.

The High Performance Computing.Communisations, and Information Technology(HPCCIT) Subcommittee and its ExecutiveCommittee coordinate Program planning. bud-geting. implementation. and review: meet withother Federal organizations with mutual inter-ests: and communicate with other Federal agen-cies and the U.S. Congress. In FY 1994, theSubcommitteemet with technology experts fromleading hardware, telecommunications, and ]oft-ware companies. The representatives wereasked to comment on the Program and provideadvice on technical and programmatic issues inhigh performance computing and communica-tions. Similar meetings are planned with otherconstituencies.

HPCCIT working groups coordinate the effortsin the various Program components. The newest

.one. the Information Infrastructure Technologyand Applications Task Group. established in1993. has developed coordinated agency plansfor R&D in NII technology and NationalChallenge applications. The networking work-ing group coordinates its activities with theFederal Networking Council (FNC) whose mem-bership extends beyond HPCC. The FNC estab-lishes strategy and policy direction, provides fur-ther coordination. and addresses technical. oper-ational. and management issues of theInteragency Internet. Its Federal NetworkingCouncil Advisory Committee represents a broadspectrum of network users and providers. Bothindividual agencies and these workima groupssponsor meetings with industry and academia.including biennial meetings with hardware ven-dors to discuss plans, and periodic GrandChallenge workshops.

The earl success of the HPCC Program is nowproviding it with one of its greatest managementchallenges. as more programs use its technolo-gies. as new sectors of the economy are relyingon scalable parallel systems. and as more peopleare becoming aware of its value to the Nation.The Program itself is highly leveraged acrossmany other governmental and private sector pro-grams. and the synergy between the emergingtechnologies and pilot applications of this R&Dprogram is revealed in numerous other applica-tion areas. Practically every Federal agency hasactivities that are not part of the HPCC Programbut that use the technologies and results from theProgram. Some of those associated with agen-cies that participate in the Program are describedin this document. Other education, health care.design for manufacturing. and related scientificresearch programs are not part of the HPCCProgram but depend on HPCC technologies.The explosive growth of computing and net-working will result in the pervasiveness of highperformance computing and communicationsthat are die foundation of the NII.

10. Information about the HPCC Program

In l'Y 1993, the NCO held some 50 meetingswith representatkes front the Federal and othergovernments. industry, and academia. The

I

6.;

33

HPCC Program actively seeks input and advicefrom all interested parties.

The NCO has provided electronic, print, andvideo materials to hundreds of media representa-tives in the U.S. and abroad, and responded tothousands of information requests fromCongressional offices, industry, academia, andthe public. The NCO distributed over 206J0copies of its FY 1994 Annual Report entitled"High Performance Computing andCommunications: Toward a NationalInformation Infrastructure." That 186-pagebook details Program goals. activities, accom-plishments. and plans, and complements thisshorter document. Some 300 copies of the six-

National Coordination Mire for HPCC

41 1..11e N1.4.4 tv.r..4-Pettnernuot CiverytIrs.1-nif .v.n.ata

Ito in,. TAN Ott lurmet ITY,I, .. 1:1%

NW art 141,

- 1 'lomut-,to- 154 P1.14

=11PCS.S4115.1X.1

1:tr.naLliaga,

141/ T.14.-4.1m....1,

lia.S.Sk13.11011104.c..11...,..1

1)..411/ t Re.r.111,tr...=11.1.....7.1

Malaulitssasai

14-0113.41.4..1),,T4Laji

CiNtalSIAMUblana

"Home Page" for the Mosaic interface to theHPCC WWW (World-Wide Web) server.Established in October 1993, this server wasaccessed by 900 users in February 1994. ItsURL (uniform resource locator) is:http://www.hpcc.gov/

Mosaic was developed at the National Center forSupercomputer Applications (NCSA), an NSFSupercomputer Center. Mosaic may be the first"killer ap," that is, an extremely successful appli-cation in the National Information Infrastructure.Thousands of servers around the world can beaccessed using Mosaic, which is available forUnix, DOS Windows, and Macintoshes.

minute video of the same name have been dis-tributed. The video has been used in a numberof telex ision reports.

In FY' 1994. the NC() established World WideWeb and Gopher servers as well as anonymousVIP (file transfer protocol) capabilities for elec-tronic dissemination of information about theIIPCC Program over the Internet. The serversallow users to view and download informationabout the HPCC Program including the FY 1994Annual Report, this FY 1995 Annual Report.information about sources of funding for HPCCR&D. and related materials. These servers alsopros ide links to other servers, including thosemanaged by HPCC agencies and the InformationInfrastructure Task Force. At Supercomputing'93, held in November in Portland. OR, the NCOdemonstrated these technologies and distributedmaterials to thousands of visitors.

34

.0M.111

High Performance Living Today andTomorrow

High performance computing and communica-tions have already had a profound impact on thelives of all Americans. Among the benefits:better prediction of natural disasters such as hur-ricanes: more rapid deployment of needed ser-vices in times of crisis: the use of more andmore powerful computers to solve complexproblems never dreamed of a decade ago.

In the past year, the concept of the "InformationSuperhighway" has captured the imagination ofpersons from all walks of life and around thecountry. Millions of Americans have discoveredits precursor. the Internet. Millions more haveread about it. Even more are vaguely aware thatcomputers can somehow help them obtain infor-mation rapidly and communicate instantaneous-ly with others around the world.

Americans are also discovering that computingand communications help to save lives, directscarce resources, and locate survivors. When anearthquake struck Southern California this win-ter. a record number of persons logg,x1 ontocommercial and nonprofit networks to learnabout the safety of loved ones.

A typical example is that of a 62-year-oldoman in Arifona who was concerned that she

could not reach her son in the Los Angeles areafollowing the quake. She called her brother nearIndianapolis who called his son in Dallas, whoin turn used his home computer to inquire aboutthe whereabout, of his cousin. Several hourslater. a total stranger in California called themother to report that her son was unhurt.

A record cold spell and snow fall paralysedtouch of the East Coast this winter. Tens ofthousands of businesses were closed. What hap-pened? Hundreds of thousands of orkers

35

stayed home. but many of these continued work-ing. As snow and ice blanketed the East Coast,they communicated electronically and telecom-muted while avoiding the icy roads. Similarlyin California after the quake. where demolishedfreeways and unstable buildings made a normalbusiness routine nearly impossible. a recordnumber of workers and businesses turned totelecommuting as the region began to rebuild.

Imagine the possibilities with a ubiquitous infor-mation 'infrastructure, one accessible by allAmerican,. which allows u....Ts to use two-wayvideo as easily as they use the telephone or faxtoday.

The Federal government and the HPCC Programfirmly believe that the Nil should and will bebuilt primarily by the private sector. One role ofgovernment in this monumental undertaking willhe to conduct government information intensiveactivities in ways that support the testing. devel-opment. and deployment of new technologies tomeet the nation's networking and computingneeds in the coming century.

The following scenarios illustrate how high per-formance computing and communications andthe NIL can benefit all Americans in the comingyears. Many of these activities are already orwill soon be taking place in research and educa-tional settings. Much of the future described isalready here. but only for a relatively smallnumber of users. The challenge faced by thegovernment is to ensure that these benefits areaccessible by all Americans from all walks oflife. And that we spur the development of thebest possible technologies for a truly NationalInformation Infrastructure to be shared eqt lyby all our citi/ens.

-... NMI11111,

CRISIS MANAGEMENT: EARTHQUAKERELIEF IN THE YEAR 2000

A series of minor tremors along a majorCalifornia fault line over the past two monthshas fueled more than premonitions that the bigone" might soon come. Real-time assessmentsof the probability and potential location of asevere earthquake have been continuouslyrelayed to state and local emergency managersvia the National Information Infri;tructure. Inturn, these managers use their real-time logisticsand management control system to determinethe state of readiness in each community withinthe region.

These last minute preparations are just in time.Within seconds of the massive earthquake. real-time computer analysis of seismic monitoringdata from a broad array of sensors placed acrossthe region and nation pinpoint the location andmagnitude of the quake. Before the initial jolt isover, an automated regional resources manage-ment system combines the new data with a largedata base that describes in detail the geologicalstructure of the area. A map appears showingthe areas most likely affected, and the kind anddegree of earth motion expected in each area.

This geographically referenced information isconsistent with a standard geographic encodingof the area, which enables other emergencyresponse systems to spring immediately intoaction.

Within the first 30 seconds of the onset of thebig one." this information is led into still anotherinterconnected computer system that contains adetailed data base of buildings. highways. andother structures in the affected area. The systemshows the expected ability of these structures tohold up after the kind of earth motion just expe-rienced and that to he expected as a result ofaftershocks from a quake of this magnitude.

Another database predicts the expected distribu-tion of people in those areas. taking into accountthe time of day. special events that are under-way, and real-time data for hotel and theateroccupancy rates, traffic distribution, and the

36

location of hazardous materials in every build-ing. truck, and train throughout the area.

Still within a minute of the initial jolt, a masterdatabase combines all this and other data, suchas the prevailing and forecasted weather condi-tions. to create an instant picture of the likelynature and extent of this major emergency.

Alarms are quickly activated in emergency con-trol centers. fire and police stations, hospitalsand other medical care centers, and in the officesand homes of local. state, and federal emergencyofficials. By the end of this first minute,"informed" emergency bulletins immediatelyinterrupt the routine in all schools, offices andhomes in the area, providing a concise messageof what has happened and what to do.

As a result, emergency resource teams aredeployed within a few minutes of the quake,knowing where they are most needed and whattypes of injuries to expect. Hazardous materialsresponse teams are dispatched to the neededsites, armed with current information about thelocation and nature of the hazardous materialsand the latest weather data. Appropriate warn-ings are issued, allowing the general public totake immediate action to protect their safety.

In the hours following the quake, this informa-tion is updated continuously and is relayed toemergency support teams, hospitals, governmentofficials, and private sector managers. Wirelessinterfaces and satellite links ensure the rapidtransmission of critical information to teamsworking in areas where phone and electricallines are disrupted. Community centers and oth-ers register millions of local residents for quicklocation by loved ones and a first step in emer-gency assistance programs.

The result? Thousands of lives saved throughfast, appropriate emergency response. and bil-lions of dollars of property protected fromunnecessary danger. such as hazardous materialsor other threats.

-.1

EDUCA'T'ION AND LIFELoNc, LEARN- craft within weeks rather than the year it used toING

Advanced computing and communications tech-nologies have revolutionized the way many stu-dents learn. Elementary school students corre-spond with electronic pen pals around the world.learning more about faraway lands. cultures andcurrent events than from mere lectures: studentsfrom three geographically distant high schoolscollaborate to simultaneously measure the dis-tance from the earth to the sun, learning invalu-able lessons in math, science, and communica-tions: other students visit distant art museumsand other educational sites via "electronic fieldtrips.- Interactive multimedia encyclopediasand other learning tools allow students to selecta myriad of information sound, video, maps,charts, and text on virtually any subject.

13ceziuse of high performance computing andcommunications, academic research will contin-ue to change at an increasingly rapid pace. Themany cAlaborations of scientists in the 1990s.such as remotely sharing scientific instrumentsand simultaneously working on scientific prob-lems from distant sites, will. within a few years.likely result in even more and increasinglysophisticated collaborations involving geograph-ically diverse researchers and institutions.

For example. a National Virtual Laboratory willallow geographically distributed researchers toshare experimental results, collaborate on teamresearch projects. and share coursework for theirstudents who are also geographically distributed.Likely collaborations will include projects thatare too expensive for one institution to perform,such as research on robotic vehicles.

Simulation-based education and training willincreasingly he used for on-the-job training. Forexample. the aircraft industry-has long usedHight simulators for pilot training, especially inhandling emergency situations. Today, simula-tion is also being used in training across the field

while aircraft are still under construction.mechanics are learning maintenance and repairvia computer simulation. By the time the planesare built, these mechanics will he able to confi-dently and efficiently maintain and repair the

37

take.

Education and the Nil il7Y1 2000

Fifteen blocks from Yankee Stadium in theSouth Bronx. it is the elective period for severalsixth grade students in P.S. 91. They are sittingin front of computers and are wearing head-phones with attached microphones.

Renaldo is studying Chinese with two dozenother students scattered around the city led by ateacher in Queens. On the screen two childrensay, in Chinese. "It's nine o'clock.- The teacherasks Renaldo to repeat the phrase. "Very good,Renaldo," she says.

At the next desk. Laverne is working on a solidwaste project with Cindy. her "key pal" inSchoharie County in upstate New York.Laverne and Cindy met when Laverne posted aquestion on a bulletin board about recycling.Cindy lives on a farm and organized a plasticrecycling program when the local dump was indanger of exceeding its capacity.

Eric is in the music section of the library assem-bling material for his research "paper'' aboutDuke Ellington. I-fe attaches pieces of a 1942audio recording and the 1959 video clip of "ATrain" to illustrate the changes in orchestrationstyles between those times.

Mrs. Esformes, the classroom teacher. is"attending" a seminar on teaching visuallyimpaired children. She has a blind student thisterm and she is comparing. her experiences andtechniques with 14 other teachers from aroundthe city who are also working with severelyvisually impaired students for the first time. Theleader of the seminar is a professor in Madison,WI.

1,ifrI011,1; Learltirtg

The NI1 will also aid in lifelong learning, mak-ing it more accessible to millions of Americans.Jennifer, a secretary at a small marketing fit in,

wants to complete her college education andreceive a degree in geology. Although the smallbusiness for which she works has no formal edu-cational assistance program. they oft-es to let heruse their computer: and network connection onher off hours. During her lunch break, forexample, Jennifer is enrolled in an interactiveclass on soil conservation. She is able not onlyto listen to and see the lecture, but can ask ques-tions via a microphone at her workstation. Ifother classes, however, are not offered at a timeshe can attend, the computer can store the trans-mission of the class for later viewing. Questionscan he mailed electronically to the professor,who can send answers hack the next day. Andwhen it is time to write a term paper on theeffects of heavy rains in the Rocky Mountains,Jennifer can electronically access databases andacademic and specialized libraries around theworld. She retrieves, electronically. a topo-graphical map of the Western Slope of theRocky Mountains. Using a mouse, she movesan icon to a certain point on the ridge. andreleases a drop of water to see which way itflows downhill Using existing databases, she isable to create a scenario for the soil effects ofheavy rain in the area, including the projectedloss of specific minerals in the land.

ELECTRONIC COMMERCE

Electronic Banking

Imagine 20 years ago. being told that you couldgo to a city virtually anywhere in the world andreceive instant cash with a wallet sized card.Not so many years ago. emergency cash on aweekend was something obtained from friendsor relatives with ready money or through thegraces of a check-cashing card at the local gro-cery store. Today. automated teller machinescan providt cash across the country and aroundthe world.

Automated teller machines have revolutionizedthe way we hank. Americans no longer try to fittheir hours around the narrow "bankers' hours"of yesteryear. And banking institutions haveactually expanded their operations in response

offering satellite banks in grocery stores, and

38

evening and even Saturday hours. Commercialelectronic shopping networks already exist: taxreturns can be filed electronically. Why notonline applications for loans or other financialneeds that can he completed at home at one'sleisure?

These are but a few examples of the many bene-fits electronic commerce can bring to Americansociety. But, unless developed carefully andwisely. these benefits could carry enormousrisks. To date. electronic commerce has takenplace in protected networks maintained by banksor other businesses. Ensuring privacy. security.and authenticity is essential for a NationalInformation Infrastructure.

Today's Internet is a fabulous example of what alittle federal investment can stimulate. But itwas designed as. and remains. a tool forresearch and education. an opportunity toexplore what is possible and what works best.HPCC-funded gigabit testhed sites explore notonly higher speed transmission of data. but howsuch data can best he used and shared in the realworld, and how best to protect the privacy of itsusers. Applications such as electronic bankingcan h from these efforts.

Eleonnli Brokerhig

Electronic brokering services are alreadydemonstrating significant savings in time andmoney. The ARPA-funded Fast Brokering sys-tem for small purchases. for example. hasproven in several pilot studies that it can operatelive times faster than standard procurement pro-cedures. When an item is needed. a data basesearch considers not only the lowest price. butthe projected delivery date, and past perfor-mance by the suppliers, to determine the mostsuitable provider or providers for the item.Potential vendors are then notified electronical-ly. or via fax or phone. and asked for pricequotes and delivery dates. Currently in pilot useat some 20 military bases across the country, thesystem has reduced the number of days front therequest for quotations to completed procurementfrom Y( to 20 days.

.F141W1111111114111111W

Under this system. intragovernment lines offunding are established with the broker, whothen transfers the funds to the supplier uponacceptance of the delivery. New payment mech-anisms, such as credit card or debit payments.will need to he integrated into the system to sup-port the more spontaneous transactions demand-ed by consumers.

Consumers could receive price quotes on specif-ic products as well as related reports on perfor,mance and customer satisfaction, and informa-tion on suppliers (S'uch as store hours, deliveryterms. warranties, length of time in business,and reports from consumer protection agenciessuch as the Better Business Bureau).

Electronic commerce will enable consumers tomake more educated choices about their pur-chases. Imo at the same time, enable businessesand manufacturers to provide the kinds of infor-mation and products most needed by Americans.

BUYING A CAR IN THE YEAR 2000

Jim and Rhonda grudgingly agree it is time tobuy a new car. They have held on to their oldvehicle in part because of how much they hatedtheir last car buying experience the hoursspent at the library evaluating specific cars: theseemingly endless trips to car dealers where theywere provided with little information other thanglossy brochures: the hassles of getting a loanand new insurance: and the disappointment uponlater learning that one of their neighbors boughtthe same car for much less.

This time. they think, maybe their computer canhelp. They know they need a larger vehicle fortheir growing family, but which is best a

minivan or station wagon'? A quick literaturesearch pulls up independent reviews that outlinethe advantages and disadvantages of each

Still not sure which they would prefer. Jim andRhonda review safety and performance reportson two minivans and two station wagons chosenfrom their initial search. The computer providesindependently produced reports on consumer

39

satisfaction, as well as the estimated mainte-nance and operating costs of each model.

The couple now decides to take a look at thecandidate cars the automobile manufacturer'scomputer server allows them to see a video ofthe car, as well as make selections to view dif-ferent colors, interiors, upholstery. and otheroptions. The interactive display allows them toselect options such as engine size, sound system.rear passenger air bags. and customized climatecontrols, while at the same time showing theestimated cost for these options and the impacton fuel consumption. This computer serviceactually allows the couple to custom design theircar, choosing power brakes, for example, whilebypassing power windows if they prefer. Theavailability of a car with the chosen options isshown at the bottom of the screen, along with anestimated delivery time for special orders.Reviewing the various packages available, thecouple narrows their choice to two models.

While Jim and Rhonda could order their cardirectly from the manufacturer, like most driversthey want to experience for themselves how thecar feels and handles, and find a reputable dealernear their home who can perform routine main-tenance and handle repairs.

Using electronic "yellow pages," they requestinformation on dealers carrying the two carslocated within a 10-mile radius of their home,who also have service hours on Saturdays anduntil at least 8 p.m. on weeknights. and offershuttle service to and from public transportationor work. These arc displayed on a computer-generated map. Using the computer mouse toclick on a specific dealer, information is dis-played on the length of time the dealer has beenin business, and other considerations, such aswhether loaner cars are offered, and customersatisfaction rates.

Jim and Rhonda electronically notify two deal-erships that they would like to make appoint-ments for test drives on Saturday morning.

Using an electronic brokering service. the cou-ple now checks information on the best adver-tised price for the cars, and whether it is best to

7-71318111111

lease or buy. The service also checks for thebest loan terms available for a new car purchase.and offers electronic loan application forms thatcan he filled out, then electronically sent to thechosen lender. Another click and the couplecompares insurance rates for the two cars is

one more likely to he stolen and therefore moreexpensive to insure? How much can they saveon insurance rates by investing in an alarm sys-tem? What company offers the best rates fortheir particular driving history and needs?Which has the greatest level of cpstomer satis-faction in processing claims?

On Saturday morning. the couple test drives thetwo cars. That afternoon. they issue an electron-ic bid, which is answered by an electronic com-merce service. They then pick the best offerand, upon acceptance. activate a process thatorders their new car from the factory.Electronically generated loan and insurance pro-cedures are carried out as well, so that a weeklater, when the dealer's shuttle conies to pickthem up at their home, the couple signs thepapers and drives home in their new car. "Thatwasn't so had now, was it?" asks Jim. "No."responds Rhonda. "I can't imagine why I dread-ed it so much. It was really quite fun."

Much of the technology described in this exam-ple of electronic commerce is available today.but on a far more limited basis. These innova-tive services will primarily he developed by pri-vate industry and public consumer organiza-tions. The Federal government's role in theseefforts will he to help develop the most effective*technologies, ensure standards for interoperabili-ty of different multimedia systems. and spur thedevelopment of security measures to ensure pri-vacy and protect against consumer fraud.

Advanced security protections such as encryp-tion technology and authentication measures arecrucial to the widespread use of electronic com-merce, and the ability for Americans to conducttheir business over the NII.

Services to support the publication:dissemina-tion. and access of multimedia information willmake electronic commerce readily available tothe American public as \' ell. Intelligent services

40

will make it easy and convenient for first-timeusers to easily browse information spaces with acombination of speech and graphics and to dele-gatetasks associated with brokering to automat-ed agents.

These services extend far beyond consumer con-venience when Jim and Rhonda access thecar manufacturer's computer. for example. ittracks the combination of features and optionsmost sought by browsers. alerting the manufac-turer to those that consumers are most and leastinterested in. A special option that is viewedfrequently. but then rejected because of theprice, could he popular enough to make in largerquantities for a lower price. These data, as wellas those on what customers actually buy. canhelp auto manufacturers respond quickly to cus-tomer needs, and at the same time, increase theircompetitiveness in the global economy.

INCREASED PRODUCTIVITY THROUGHIMPROVED ENVIRONMENTAL DATA

The use of the latest information technology.coupled with more timely and complete environ-mental information provided by the government.is essential for U.S. industry to succeed in thehighly competitive global marketplace.

American industry already makes substantial useof real-time and historical weather and otherenvironmental information from NOAA to opti-mally route airplanes, trucks, and ships, thei-ebyminimizing transit time and fuel consumptionand assuring safer travel.

More timely access to more comprehensiveenvironmental information will support bothday-to-day and long-term business decisions bythousands of large and small. companies.Marketing and distribution decisions, key ingre-dients in economic growth. are driven by readilyavailable information. Companies are already.using high performance computing technologyri:',iieered by the HPCC Program to sort throughextremely large volumes of data, seeking pat-terns for potential sales of products and ser icesthat will both minimize resource consumptionand maximize bottom lines. Increased availabil-

ity of the Nation's vast amounts of environmen-tal data on the Nil will allow companies toimprove the ways their inventory and distribu-tion operations respond to changing weatherconditions across the nation. This increasesproductivity and serves customers better.Weather sensitive indicators can he better fac-tored into product development and marketingactivities to achieve additional. increases in pro-ductivity.

ENVIRONMENTAL MONITORING INTHE YEAR 2000

Michele. a biological oceanographer and fishingindustry analyst. sits at her workstation at theWestern Regional Environmental Center analyz-ing predictions for a record pollock catch offKodiak Island. The source? Observations froman on-going cruise in Alaska's Shelikof Straitthat suggest the recent El Nino-SouthernOscillation (ENSO) event is influencing the fishFur est in the area.

The analyst. whose job it is to guide the U.S.fishing fleet, has her doubts. however. She isaccessing data from satellite-linked moored sub-surface ocean sensors. polar-orbiting satellites.numerical forecasting models, and coastal sam-pling stations from the past three months. Allthese data indicate that warm tropical water hasbeen progressively moving northward along theU.S. and Canadian Pacific coast. Real-time datafrom satellite-linked drifting buoys and currentmeter arrays. as well as ship surveys of fish eggand larvae. suggest a major circulation anomalyis occurring in the spawning region off the coast.In fact, what looked like a high catch year seemsto he turning into an economic bust!

Observations from an on-going cruise in thearea, as well as the beginning of disturbingreports from fishing boats also hint that some-thing may he wrong.

Using a mouse. Michele clicks a satellite iconthat do nloads the most recent 24-hour set of

isible and infrared imagery. from NOAA andNASA ocean-observing satellites. Alter click-ing another icon. a new indow opens that

41

allows her to extract and compile a time seriesof the last 30 days of infrared imagery from theNASA satellite database in Pasadena, CA, andthe NOAA satellite database in Suitland, MD.Using tools available on the NII, she is able toquickly integrate these disparate data sets spreadacross the country and to create an animationthat clearly shows the progression of warm trop-ical waters toward the Alaskan coastal regionthat serves as a safe nursery for pollock larvae.

In order to put this environmental event into per-spective. Michele next queries the NationalOceanographic Data Center ship observationsdata base in Washington. DC. for the past 50years of sea surface temperature data. Using agraphical user interface available on the Nil thatlets scientists intuitively explore and visualize avariety of multi-dimensional data products, shezooms in on the North Pacific and quickly cre-ates a 50-year animation loop of sea surfacetemperature. In another window of her worksta-tion, she accesses the ENSO and EquatorialUndercurrent oceanographic database at theUniversity of Washington. Next she creates anew graphical overlay of her data that suggests along-term, phased relationship between equato-rial processes and sea surface temperature in thenorthern Pacific. This visual observation is con-firmed by running correlation and coherenceanalyses using a point-and-click time seriespackage, available on the Nll from the ScrippsInstitution of Oceanography in La Jolla. CA.

Based on these data, she is able to project near-by regions where the fish are likely to go. Withanother icon selection, she overlays severaltracks of drifting buoy data and subsurfacemoored current meter data onto the satelliteimagery. These data suggest that th, plannedtrack of an observation cruise underway willhave to he modified to properly sample the newcirculation feature and the whereabouts of thefish. To plan for the new cruise track, shechecks the online National Weather Service'sfic -day weather forecast and the Nay's PacificOcean Circulation Model forecast for the studyregion.

The ship's captain is fully informed thnigh theship's onboard workstation, which is networked

..... .11

by satellite link to the shore-based support sys-tems. of the relevant observational and environ-mental forecast information. He instantlyaccesses electronic maps and charts to plan achanged course. Several hours later comes thenews yes. the fish have moved. Michelenotifies the Northwest Pacific Fishery Council,which in turn issues a recommendation to relo-cate the commercial fishing activities.

DELIVERING HEALTH CARE TOREMOTE AREAS

Lisa knew something was seriously wrong whenthe pain awakened her from sleep. She was nostranger to adversity, having lived her life in theAppalachian highlands where getting help withany sort of problem meant a long drive overwinding mountain roads. But being a first-timemother -to -he brought special challenges, andthis throbbing headache the worst she'd everexperienced made her sense of isolation asdark and deep as the night's sky over the BlueRidge peaks. She picked up the phone andcalled the emergency number at the ValleyHealth Clinic. The nurse practitioner whoanswered the phone told her what she alreadyknew: Dr. Clark, one of a few circuit-ridingobstetricians who covered this sparsely populat-ed part of the state. would he holding the com-ing day's clinic hundreds of miles away. Butthe nurse added, "We'll he able to get his help asif he were right here." and directed Lisa's hus-band. Mike, to get her to the clinic promptly.

The couple arrived at the little clinic just as firstlight was owl:ning the nearby ridges. Lisa wasworse, her vision becoming occasionallyblurred. She knew the baby wasa't due foranother four weeks. and she was scared thatthese symptoms might mean the baby wasdying. The comforting words of the nurse. andher careful exam. helped to lessen the fear. butthe findings were not good: Lisa's blood pies -sure was elevated and the fetal heart soundswere weak. The nurse said it might he a diseasecalled "toxemia of pregnancy" and that shemight need to go to the hospital. "Hospital! Butthat's a four-hour drive over the mountains. andif she could get better just by resting, we sure

42

don't want to put her through that." Mikereplied.

A few years before, the nurse and the familywould have had to call Dr. Clark on the phoneand get his opinion, an educated guess really.about whether to undertake the arduous journey.But high performance computing and communi-cations had changed that. For the telemedicineworkstation in this little two-room clinic nowbrought medical expertise from far away right toLisa's side. Initiating a two-way video link, thenurse brought Lisa "face-to face" with the doc-tor. who had been reviewing the previous day'sX-rays at his own office workstation severalhundred miles away. He greeted Lisa andobserved from her expression that she was inmuch pain. He asked a few questions about theheadache and vision problems. then asked thenurse to put the fetal ultrasound transducer onLisa's abdomen. The image of the baby in thewomb simultaneously appeared on the clinicworkstation screen and on the distant physician'sworkstation. Dr. Clark guided the nurse to turnthe sound beam right and left to obtain a betterview as nurse, doctor, and patient watched thescreen. The fetal electrocardiogram, a continu-ous tracing of the baby's hcart, traced a regularpattern in another window on the workstationwhile Lisa's blood pressure was recorded anddisplayed digitally by an automatic blood pres-sure cuff.

"This certainly looks like toxemia." Dr. Clarksaid. "hut something else is going on as wellthe baby's heart rate is intermittently slowing forsome reason. Try moving the ultrasound probeover here." Using an on-screen marker, hepointed out the baby'S head and neck, and thenurse tilted the pencil-like probe that restedlightly on Lisa's abdomen. "There it is!" thedoctor exclaimed, as the outline revealed theumbilical cord wrapper several times around thebaby's neck. As the child moved in the womb,the umbilical blood supply was occasionallypinched and the baby's heart rate slowed down.a sign of fetal distress.

Lisa would need a Cesarean section to preset:\ ethe baby's health. What might have been aneducated guess and an unexplained abnormality

-before high performance computing and com-munications was instead a certain diagnosis.And the hospital caring for her was well pre-pared for immediate action upon her arrival.The nurse used multimedia privacy-enhancedelectronic mail to send the ultrasound images.electrocardiograms. and even Dr. Clark's videointeraction with the patient to the receiving hos-pital staff. A simple point-and-click interactionon the workstation screen gathered the varioussignals and images. Together with copies ofLisa's prenatal care clinic notes. the nurse sent acomprehensive electronic patient record to thedistant facility. When Lisa arrived later thatday, the ward staff recognized her immediatelyand welcomed her as if she were an old friend.

The baby girl delivered by Cesarean section thatafternoon started life a few weeks early. butgrew and thrived. it would he many yearsbefore she could comprehend or even spell"National Information Infrastructure." but hermom would say. "We had to save your life whenyou were just a picture on the screen." The deci-sions made with speed and certainty that fatefulmorning depended on many information tech-nologies: ubiquitous high speed digital commu-nications techniques: medical computer work-stations with advanced motion graphics andideo: and secure methods to protect confiden-

tiality and privacy of network communications.This young mother in rural Appalachia stilldoesn't know much about computers. but sheknows she got speedy and effective health care.And she knows the bright eyes of a health)child, brought safely into the world with the helpof high performance computing and communi-cations.

Monitoring Hurricane Andrew from geostationary orbit. This image from the NOAA GOES-7 spacecraft isan example of the type of data that are being made available on the Internet and are of great interest to thepublic as well as the meteorological community.

44 51.

BEST COPY AVAILABLE

HPCC Program Summary

HPCC Program Goals

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 the national economic competitiveness,

national security, education, health care, and the global environment.

Provide key parts of the foundation for the National InformationInfrastructure (NI1) and demonstrate selected NII applications.

HPCC Agencies

ARPA - Advanced Research Projects Agency, Department of Defense

DOE - Department of Energy

ED - Department of Education

EPA - Environmental Protection Agency

NASA - National Aeronautics and Space Administration

NIH - National Institutes of Health, Department of Health and Human

Services

NIST - National Institute of Standards and Technology, Department of

Commerce

NOAA - National Oceanic and Atmospheric Administration, Department of

Commerce

NSA - National Security Agency, Department of Defense

NSF - National Science I. oundation

HPCC Program Strategies

Develop, through industrial collaboration, high performance computing sys-tems ust.:g scalable parallel designs and technologies capable ofsustaining atleast one trillion operations per second (teraops) performance on large scien-tific and engineering problems such as Grand Challenges.

Support all HPCC components by helping to expand and upgrade theInternet.

Develop the networking technology required for deployment of nationwidegigabit speed networks through collaboration with industry.

Demonstrate the productiveness of wide area gigabit networking to supportand enhance Grand Challenge applications collaborations.

Demonstrate prototype solutions of Grand Challenge problems that achieveand exploit teraops performance.

Provide and encourage innovation in the use of high performance computingsystems and network access technologies for solving Grand Challenge andother applications by establishing collaborations to provide and improveemerging software and algorithms.

Create an infrastructure, including high performance computingresearchcenters, networks, and collaborations that encourage the diffusion and use ofhigh performance computing and communications technologies in U.S.research and industrial applications.

Work with industry to develop information infrastructure technology to sup-port the National Information Infrastructure.

Leverage the HPCC investment by working with industry to imph,mentNational Challenge applications.

Enhance computational science as a widely recognized disciplinefor basicresearch by establishing nationally recognized and accepted educational pro-grams in computational science at the pre-college, undergraduate, and post-graduate levels.

Increase the number of graduate and postdoctoral fellowships in computerscience, computer engineering, computatioual science and engineering, andinformatics, and initiate undergraduate computational sciences scholarshipsand fellowships.

46

Overview of the Five HPCC Components

Five integrated components represent the key areas of high performance computingand communications:

HPCS High Performance Computing Systems

Extend U.S. technological leadership in high performance computing through thedevelopment of scalable computing systems, with associated software, capable of sus-taining at least one trillion operations per second (teraops) performance. Scalable par-allel and distributed computing systems will be able to support the full range of usagefrom workstations through the largest scale highest performance systems. Workstationswill extend into portable wireless interfaces as technology advances.

NREN National Research and Education Network

Extend U.S. technological leadership in computer communications by a program Ofresearch and development that advances the leading edge of networking technologyand services. NREN will widen the research and education community's network con-nectivity to high performance computing and research centers and to electronic infor-mation resources and libraries. This will accelerate the development and deploymentof networking technologies by the telecommunications industry. It includes nationwideprototypes fur terrestrial, satellite, wireless, and wireline communications systems,including fiber optics, with common protocol support and applications interfaces.

ASTA Advanced Software Technology and Algorithms

Demonstrate prototype solutions to Grand Challenge problems through the develop-ment of advanced algorithms and software and the use of HPCC resources. GrandChallenge problenzs are computationally intensive problems such as forecasting weath-er, predicting climate, improving environmental monitoring, building more energy-effi-cient cars and airplanes, designing better drugs, and conducting basic scientificresearch.

HTA Information Infrastructure Technology and Applications

Demonstrate ,irototype solutions to National Challenge problems using HPCC enablingtechnologies. .tional Challenges are infarmationally intensive applications such aseducation and lifelong learning, digital libraries, health care, advanced manufacturing,electronic commerce, and environmental monitoring. II7'A will support work tograte technologies, such as services, software, and interfaces. to bring HPCC benefitsto the general public. These will be leveraged across the National Challenges, leadingto significant economies of scale in the development costs.

BRHR Basic Reseak-h and Human Resources

Support research, training, and education in computer Sc ience, computer engineering.and computational science, and enhance the infrastructure through the addition ofHPCC resources. Initiation of pilot projects for K-I2 and lifelong learning will supportexpansion of the NIL

1,

47

3

!.;;L:24-1,11NIIIMNIMMIPP

Evaluation Criteria for Agency Participation in the HPCC Program

Relevance/Contribution. The research must significantly contribute to theoverall goals and strategy of the Federal High Performance Computing andCommunications (HPCC) Program, including computing, software, networking,

information infrastructure, and basic research, to enable solution of the GrandChallenges and the National Challenges.

Technical/Scientific Merit. The proposed agency program must be technical-ly/scientifically sound and of high quality, and must be the product of a docu-

mented technical/scientific planning and review process.

Readiness. A clear agency planning process must be evident, and the organiza-tion must have demonstrated capability to earn' out the program.

Timeliness. The proposed work must be technically/scientifically timely for oneor more of the HPCC Program components.

Linkages. The responsible organization must have established policies, pro -grans, and activities promoting effective technical and scientific connectionsamong government, industry, and academic sectors.

Costs. The identified resources must be adequate, represent an appropriateshare of the total available HPCC resources (e.g., a balance among programcomponents), promote prospects for joint funding, and address long-termresource implications.

Agency Approval. The proposed program or activity must have policy-levelapproval by the submitting agency.

48

Age

ncy

Res

pons

ibili

ties A

RP

AN

SF

DO

E N

AS

AN

IHN

IST

NS

A N

OA

A E

PA

ED

Hig

h P

erfo

rman

ce C

ompu

ting

Sys

tem

s (H

PC

S)

IR

esea

rch

for

Fut

ure

Gen

erat

ions

of C

ompu

ting

Sys

tem

sX

XX

XX

2 S

yste

m D

esig

n T

ools

XX

X

3 A

dvan

ced

Pro

toty

pe S

yste

ms

XX

XX

X4

Eva

luat

ion

of E

arly

Sys

tem

sX

XX

XX

X,

Nat

iona

l Res

earc

h an

d E

duca

tion

Net

wor

k (N

RE

N)

1In

tera

genc

y In

tern

etX

XX

XX

XX

XX

2G

igab

it R

esea

rch

and

Dev

elop

men

tX

XX

XX

XX

Adv

ance

d S

oftw

are

Tec

hnol

ogy

and

Alg

orith

ms

IAS

TA

I

1S

uppo

rt fo

r G

rand

Cha

lleng

esX

XX

XX

XX

X2

Sof

twar

e C

ompo

nent

s an

d T

ools

XX

XX

XX

X(I

D3

Com

puta

tiona

l Tec

hniq

ues

XX

XX

XX

X4

Hig

h P

erfo

rman

ce C

ompu

ting

Res

earc

h C

ente

rs (

HP

CR

Cs)

XX

XX

XX

X

Into

rmak

on In

fras

truc

ture

Tec

hnol

ogy

and

App

Irca

tions

;IIT

A)

IIn

form

atio

n in

fras

truc

ture

Ser

vice

sX

XX

XX

XX

2S

yste

ms

Dev

elop

men

t and

Sup

port

Env

ironm

ent

XX

XX

3In

telli

gent

Inte

rfac

esX

XX

XX

4N

atro

na! C

halle

nges

XX

XX

XX

XX

Bas

ic R

esea

rch

and

Hum

an R

esou

rces

(B

RH

R)

1B

ar-a

c R

esea

rch

XX

XX

XX

2R

esea

rch

Par

ticip

atio

n an

d T

rain

ing

XX

XX

X.1

torr

a,tr

ucrig

eX

XX

4E

duca

tron

Tra

inin

g an

d C

urric

ulum

XX

XX

XX

i11

5.7

Agency Budgets by HPCC ProgramComponents

FY 1994 Budget (Dollars in Millions)

Agency HPCS NREN ASTA IITA BRHR TOTAL

ARPA 103.9 48.7 32.3 95.6 18.4 298.9NSF 19.8 47.9 119.3 19.0 61.0 267.0

DOE 10.8 16.6 73.8 0.3 20.7 122.2

NASA 11.5 13.2 73.1 12.0 3.2 113.0

NIH 3.4 3.0 26.1 13.6 11.7 57.8

NISI' 0.3 1.9 0.6 15.3 18.1

NSA 21.5 7.1 11.2 0.2 0.2 40.2NOAA 1.0 9.8 10.8

EPA 0.7ED 2.0

5.9 1.3 7.92.0

TOTAL 171.2 142.1 352.1 156.0 116.5 937.9

FY 1995 Budget Request (Dollars in Millions)

Agency HPCS NREN ASTA IITA BRHR TOTAL

ARPA 110.7 61.1 29.6 140.8 15.2 357.4

NSF 21.7 52.9 141.2 50.6 61.1 328.6

DOE 10.9 .16.8 75.5 1.2 21.0 125.4

NASA 9.7 12.7 81.2 17.5 3.8 124.9

NIH 4.9 8.4 23.8 29.1 15.6 81.8

NIST 6.8 3.7 4.6 41.4 56.5

NSA 16.1 11.8 11.8 0.2 0.2 40.1

NOAA 8.7 . 16.1 0.5 25.3

EPA 0.7 11.7 0.3 2.0 14.7

TOTAL 180.8 176.8 395.5 281.6 120.0 1,154.7

: y 50

HPCCIT Reporting Structure andSubcommittee Roster

Office of Science and Technology ARPA

Policy (OSTP)

Director: John H. Gibbons

National Science and TechnologyCouncil (NSTC)

NSTC Secretariat

Angela Phillips Diaz. Executive Director

Committee on Information andCommunication (CIC)

Chairs: Anita K. JonesLionel S. Johns

Vice Chair: Melvyn Ciment

High Performance C imputing,Communications, andInformation TechnologySubcommittee (HPCCIT)

('hairDonald A. B. Lindberg

('harles R. Ka lina

RepresentativeJohn C. Toole

AitertunesStephen L. SquiresRandy Katz

NSF

RepresentativeMelvyn Ciment

AlternatesMerrell Patrick

Robert G. Voigt

DOE

Reproentan reDavid Nelson

AlternatesJohn S. Cava MillNorman H. Kreisman

NASA

Re/v.v./nativeI.ee 13. Holcomb

AltentatePaul H. SmithPaul E. Hunter

NIH EPA

Repre AentativeDaniel R. Masys

AlternatesJudith L. Vaitukaiti\Robert L. Martino

NSA

Repw.sentativeGeorge R. Cotter

A/ternateNorman S. Glick

NOAA

RepresentativeThomas N. Pyke, Jr.

A/ternateErnest J. Daddio

NIST

James H. Burrows

AlternateR. J. (Jerry) Linn

ED

RepresentativeLinda Roberts

AlternateJames A. Mitchell

52

RepresentativeJoan H. Novak

AlternateRobin L. Dennis

OMB

Steven IsakowitzBruce McConnell

OSTP

Michael R. Nelson

;Votes: The Advanced Rewarch ProjectsAgency. the National Science Foundation, theDepartment and Energy, and the NationalAeronautics and Space Administration holdpermanent positions on the HPCCITExecutive Committee. Two other positionrotate among the other agencies: in FY 1994these were NIST and NOAA.

Reinvsentialves of the Agency.for Health CarePolicy Research. the Centers for DiseaseContra! and Prevention. the Indian HealthService, maul the Food and DrugAdministration of the Public. Health Service ofthe Department of Human Health andServices: the Department of Agriculture: theDepartment of l'eterans Affairs: and theNational Communication.% System attend theHPCCIT Subcommittee meeting.% asobservers.

Glossary

ACTSAdvanced CommunicationsTechnology Satellite, a NASA-spon-sored program. A joint NASA/ARPANREN collaboration will demonstratehigh speed ATM/SONET transmissionover the ACTS satellite. and will pro-vide interface and operations experi-ence in mating high speed terrestrialcommunications systems with highspeed satellite communications sys-tems.

AlgorithmA procedure designed to solve a prob-lem. Scientific computing programscontain algorithms.

ARPAAdvanced Research Projects Agency.part of DOD

ASTAAdvanced Software Technology andAlgorithms, a component of the HPCCProgram

ATMAsynchronous Transfer Mode, a newtelecommunications technology, alsoknown as cell switching, which isbased on 53 byte cells

Backbone NetworkA high capacity electronic trunk con-necting lower capacity networks, e.g..NSFNET backbone

BitAcronym for binary digit

BRHRBasic Research and Human Resources.a component of the HPCC Program

ByteA group of adjacent binary digits oper-ated upon as a unit (usually connotes agroup of eight hits)

CERTComputer Emergency Response Team

Computer EngineeringThe creative application of engineeringprinciples and methods to the designand development of hardware and soft-ware systems

Computer ScienceThe systematic study of computingsystems and computation. The body ofknowledge resulting from this disci-pline contains theories for understand-ing computing systems and methods:design methodology, algorithms. andtools: methods for the testing of con-cepts: methods of analysis and verifica-tion: and knowledge representation andimplementation.

Computational Science andEngineeringThe systematic application of comput-ing systems and computational solutiontechniques to mathematical models for-mulated to describe and simulate phe-nomena of scientific and engineeringinterest

CICCommittee on Information andCommunication, part of NSTC

DOCDepartment of Commerce

DODDepartment of Defense

-611:1=11.-pam

DOEDepartment of Energy

EDDepartment of Education

EPAEnvironmental Protection Agency

ESnetEnergy Sciences Network

FDD1Fiber Distributed Data Interface

FIRSTForum of Incidence Response andSecurity Teams

FlopsAcronym for floating point operationsper second. The term "floating point"refers to that format of numbers that ismost commonly used for scientific cal-culation. Flops is used as a measure ofa computing system's speed of per-forming basic arithmetic operationssuch as adding. subtracting. multiply-ing. or dividing two numlrs.

FNCFederal Networking Council

Giga-ID" or billions of ... (e.g.. gigabits.gigatlops. gigaops)

GOSIPGovernment Open SystemsInterconnection Profile

Grand ChallengeA fundamental problem in science andengineering, with broad economic andscientific impact, whose solution canhe advanced by applying high perfor-mance computing techniques andresources

Heterogeneous systemA distributed system that containsmore than one kind of computer

--.7ila1211111111111111111111111111111111EIMININIMIL,

54

High performance computingCovers the full range of advanced com-puting systems including workstations.networks of workstations with servers.scalable parallel systems. vector paral-lel systems, and more specialized sys-tems. Scalable input/output interfaces.mass storage systems. and archivalstorage are components of these sys-tems. Included also are system soft-ware and software development envi-ronments that enable users to viewtheir workstations and the rest of theircomputing environments as a unifiedsystem.

HiPPIHigh Performance Parallel Interface

HHSDepartment of Health and HumanServices

HPCCHigh Performance Computing andCommunications

HPCCITHigh Performance Computing.Communications, and InformationTechnology Subcommittee, part of theCIC

HPCSHigh Performance ComputingSystems. a component of the HPCCProgram

IITAInformation Infrastructure Technologyand Applications, a component of theHPCC Program

Interagency InternetThe federally funded part of theInternet that is directly used by theHPCC Program. It includes NM:NET.ESnet. and NSI.

InternetThe global collection of interconnect-

ed. nu ltiprotocol computer networksincluding Federal. mid-level. private.and international networks. TheInteragency Internet is a part of theInternet.

InterNICInternet Network Information Center

IntereonneetivityThe ability of two or more computersto easily locate and communicate witheach other over an infrastructure thatprovides the speed and clarity toaccomplish a proposed task

lnteroperabilityThe ability of any two computers thatare interconnected to understand eachother and perform mutuall_ supportivetasks such as client/server computing

ISDNIntegrated Services Digital Network

Kb/sKilobits per second or thousands of bitsper second

LANLocal Area Network

Nlb/sMegabits per second or millions of hitsper second

NIINIDMultiple Input Multiple Data

NI PP

Massively parallel processor

NAPsNetwork Access Points, a set of nodesconnecting mid-level or regional net-

ork s and other service providers toWit:NET

NASANational Aeronautics and SpaceAdministration

55

National ChallengeA fundamental application that hasbroad and direct impact on the Nation'scompetitiveness and the well-being ofits citizens, and that can benefit fromthe application of HPCC technologyand resources

National Information Infrastructure(NII)The integration of hardware, software.and skills that will make it easy andaffordable to connect people with eachother, with computers. and with a vastarray of services and informationresources.

NCONational Coordination Office for HighPerformance Computing andCommunications

NetworkComputer communications technolo-gies that link multiple computers toshare information and resources acrossgeographically dispersed locations

NIHNational Institutes of Health, part ofHHS

NIS'l'National Institute of Standards andTechnology. part of DOC

NI,NINational Library of Medicine, part ofNIH

NOAHNational Oceanic and AtmosphericAdministration, part of DOC'

NRENNational Research and EducationNekvork. The NREN is the realisationof an interconnected gigabit computernetwork system devoted to HPCC.NREN is also a component of theHPCC Program.

NSANational Security Agency. part of DOD

NSFNational Science Foundation

NSFNETNSF computer network

NSINASA Science Internet

NSWNational Science and TechnologyCouncil

OMBOffice of Management and Budget

OpsAcronym for operations per second.Ops is used as a rating of the speed ofcomputer systems and components. In

this report ops is generally taken tomean the usual integer or floating pointoperations depending on what function-al units are included in a particular sys-tem configuration.

OSIOpen Systems Interconnection

OSTPWhite House Office of Science andTechnology Policy

Parallel processingSimultaneous processing by more thanone processing unit on a single applica-tion

Peta-1 O'S of ... petabits)

PortTransport a computer program Fromone computer system to another

PortablePortable computer programs can he runwith little or no change on many kindsof computer systems.

PrototypeThe original demonstration model ofwhat is expected to be a series of sys-tems. Prototypes are used to prove fea-sibility, but often are not as efficient orwell-designed as later production mod-els.

R&DResearch and development

ScalableA system is scalable if it can he madeto have more (or less) computationalpower by configuring it with a larger(or smaller) number of processors.amount of memory, interconnectionbandwidth. input/output bandwidth.and amount of mass storage.

SIMDSingle Instruction Multiple Data

SONETSynchronous Optical Network

Tera-10'2 or trillions of ... (e.g.. terabits. ter-aflops)

T1Network transmission of a DS I format-ted digital signal at a rate of 1.5 Mh/s

13Network transmission of a DS3 format-ted digital signal at a rate of 45 Mb/s

vBNSVery high speed Backbone NetworkServices

56

Contacts

NCO(National Coordination Office forHigh Performance Computing andCommunications)Bldg. 38A/BI-N308600 Rockville PikeBethesda, MD 20894301-402-4100fax: 301-402-4080nco@hpcc.gov

Donald A. B. Lindberg. M.D.Directorlindberg(?hpcc.gov

Patricia R. CarsonE.vecutive Assistantcarson@ hpcc.gov

Sally E. Howe. Ph.D.Assistant Director for Technical Pwgramshow e @ h pec.gor

David R. HowellProgram AssociateNASA Goddard Space Flight Center

1994 SESCDPhowel I OP hpcc.go\

Charles R. KalinaExecutive Secretary, HPCCIT Subcommitteekalina@hpcc.gov

James R. McGraw. Ph.D.Program AssociateDOE Lawrence Livermore National Laboratory

1993-1994 IPAmcgraw@shpec.gov

Shannon N. UncangcoStalfAssistantuncangeoCcf hpcc.gov

Patricia N. WilliamsInformation ()Meerpw illiams (6s hpcc.gov

,

HPCC Internet Servers

gopher.hpcc.gov

ass .hpec.gov

Mosaic/Vv'orld-Wide Web URL (uniformresource locator): http://www.hpcc.gov/

ARPA

Lt Col John C. TooleActing Director. Computing SystemsTechrudagy OfficeAdvanced Research Projects Agency3701 N. Fairfax Dr.Arlington. VA 22203703-696-2264toole@arpa.mil

Stephen L. SquiresSpecial Assistant to the DirectorAdvanced Research Projects Agency3701 N. Fairfax Dr.Arlington, VA 22203703-696-2226squires@arpa.mil

Randy Katz. Ph.D.Program Manager for Iniorrnationhtfrast nu:fur(' Technolog\Advanced Research Projects Agency3701 N. Fairfax Dr.Arlington. VA 22203703-696-2228katz arpa.mil

Col John Silva. M.D.Program Manager.lar Health Care InformationInfrastructureAdvanced Research Projects Agency37(11 N. Fairfax Dr.Arlington. VA 22203703-696-2221jsilvaO'arpa.mil

..11011I

DOE

HPCC Program CoordinatorOffice of Scientific ComputingER-30, GTN

DeParimeni of blergYWashington. DC 20585301-903-5800

hpcc(crer.doe.gov

ED

Linda G. Roberts. Ed. D.Specie,/ Advisor on Educational TechnologyDepartment of EducationF0136 4015Washington. DC 20202202-401-1444linda_robertsqf cd.gov

Alexis T. PoliakoffNetwork Planning Stall'Department of EducationOHRA/IRMSROB 4656Washington. DC 20202202-708-5210alex_poliakoff0'ed.gov

James A. Mitchell. Ph.D.Special As.sistamPr Ter/me/ogy011ice of the Assistant SecretaryDepartment of Education555 Nev \\ Jersey Ave.. NW #60413Washington. DC 20208202-219-2053jmitchel inet.cd.gov

EPA

General information:

Joan II. NO\ akIPCC Prov.ain Manager. MD-S0

Enril'011inental Protection AgencyResearch Triangle Park. NC 27711919-541-4545novak jowl Ca'eptnail.epa.go \

58

Robin L. Dennis. Ph.D.Senior Science Program ManagerEnvironmental Protection AgencyResearch Triangle Park, NC 27711919-541-2870rdennis@tlyer.ncsc.org

Internet use:

Bruce P. AlmichEPA Federal Network CouncilRepresentative, MD-90Environmental Protection A t;eney

Research Triangle Park. NC 27711919-541-3306almich.bruce@epamail.epa.gov

NASA

Lee B. HolcombDirector. High Performance Computing andCommunications OfficeOffice of AeronauticsNcuianul Aeronautics and Space AdministrationCode RCWashington. D.C. 20546202-358-2747I_holcomhC aeromail.hq.nasa.gov

Paul H. Smith. Ph.D.Program Manager, HPCCNational Aeronautic s and Space AdministrationCode RCWashington. D.C. 20546202-358-4617ph_smith(g'acromail.hq.nasa.gov

Paul E. FlouterPmgram Manager. //TANational Aeronautics and Space ildminiAlrafi'mCode RCWashington. D.C. 20546202-358-4618p_ Ii ter(a' acromai 1 .hq .n asa.gov

NIH

NIH HPCC Coordinator and NLM contact:

Daniel R. Masys, M.D.Director, Lister Hill Natio,. I Centerfor Biomedical Communications

National Library qt. MedicineBldg. 38At7N707Bethesda. MD 20894301-496-4441masys0P1hc.nlm.nih.gov

NC RR program:

Judith L. Vaitukaitis, M.D.Director. National Center for ResearchResourcesNational Institutes of HealthBldg. 12A/4007Bethesda, MD 20892301-496-5793judvyc /rnihn12u.hitnet @n`cu.nih.gov

Caroline Holloway, Ph.D.Director, Office of Science PolicyNational Center fOr ResearchResourcesNational Institutos of HealthBldg. I 2A/4057Bethesda. MD 20892301-496-2992CarolineH (ft nihrr12a.bitnet

DCRT program:

David Rodbard.1\1.1/Director. Division of ComputerResearch and TechnologyNational Institutes of HealthBldg. 12A/3033Bethesda. MD 20892301-496-5703rodbard nih.go

VvIlliam L. RissoDeputy Director. Division of ComputerResearch and TechnologyNational Institutes of HealthBldg. I 2A/3033Bethesda. MD 20892301-496-8277risso@nih.gov

Robert L. Martino, Ph.D.Chief Computational Bioseience andEngineering LaboratoryDivision of Computer Research andTechnologyNational Institutes of HealthBldg. I2A/2033Bethesda, MD 20892301-496-1 I I 1

mart ino@alw.nih.gov

NC1/BSC program:

Jacob V. Maizel, Ph.D.Biomedical Supercomputer CenterNational Cancer InstituteFrederick Cancer Research Facility301-846-5532jmaizel(Ltneircrigov

NIST

James H. BurrowsDirector, 'Computer Systems LaboratoryNational Institute of Standards andTechnologyTechnology Building. Room B 154Gaithersburg. MD 20899301-975-2822burrowsVinicinist.gov

...1.11111=0111

J. (Jerry) LinnAs.% wiate Director. Computer SystemsLaboratoryNational Institute of Stmulards andTechnologyTechnology Building. Room B 164Gaithersburg, MD 20899301-975-3624linnd@osi.ncsl.nist.gov

NOAA

General information:

Thomas N. Peke. Jr.Director fin. High PerformanceComputing and CommunicationsNational Oceanic and AtmosphericAdministration13 I 5 East-West Highway.Room 15300National Oceanic and AtmosphericAdministration

. Silver Spring. MD 20910301-713-3573tpyke@hpcc.noaa.gov

Directors of high performance computing cen-ters:

Jerry Mahlman. Ph.D.Director. Geophysical Fluid DynamicsLaboratoryNational Oceanic and AtmosphericAdministrationP.O. Box 308Princeton, NJ 08542609-452-6502jmOt gfdl.gov

Ron MacPherson. Ph.D.Director. Nathmal MeteorologicalCenterNational Oceanic and AtmasphericAdministration520(1 Auth Road. Room 101

60

Camp Springs. MD 20746301-763-8016ronmcp CO' 1 I c.1b4.noatt.gov

Sandy MacDonald. Ph.D.Director. Forecasting SystemsLaboratoryNational Oceanic and AtmosphericAdministration325 BroadwayBoulder. CO 80303303-497-6378macdonald@fsl.noaa.gov

Internet Use:

Robert L. MairsChief information Processing DivisionOffice of Satellite Data Processing andDistributionNational Oceanic and AtmosphericAdministrationFB4 - Room 0301Suitland and Silver Hill RoadsSuitland. MD 20746301-763-5687rmairsq.Pnesdis.noaa.gov

NSA

George R. CotterChief ScientistNational Security Agency9800 Savage Rd.Ft. Meade. MD 20755-6000301-688-6434grcotte@afterlife.ncsc.mil

Norman S. GlickSenior Computer ScientistAla tional ty A g en Cy

9800 Sto age Rd.Ft. Meade. MD 20755-600(1301-688-8448nsglick 0' a fterl i le .ncsc.mil

..NSF

n ()mem. Ph.D.Acting AsAistant Director, Directoane for('onipuler and Information Science andrnmineering\'ational Science oundanonDirectorate for Computer and InformationScience and Engineering4201 Wilson Blvd.. Room 1105Arlington. VA 22230703-306-190(1mciment(a'llsigOV

Merrell Patrick. Ph.D.i/PCC CoordinatorNational :Science FoundationDirectorate for Computer and InformationScience and Engineering4201 Wilson Blvd.. Room 1105Arlington. VA 22230703-306-1900mpat rick (P`nsf.gov

Y.T. Chien. Ph.D.//L1 :Icily/tic%National Science FoundationDirectorate for Computer and InformationScience and Engineering4201 Wilson Blvd. Room 1115Arlington. VA 22230703-306-1930

ichjen (0' nsf.go

Robert G. Voigt. Ph.D.,National Science FoundationDirectorate for Computer and InformationScience and Engineering4201 Wilson.Blvd. Room 1105Arlington. VA 22230703-306-1900r oigt Vensf.go

Peter Ariherger. Ph.D.Nanomd.Vience FoundationDirectorate for Biological Sciences4201 Wilson Bled. Room 615Arlington. VA 22230703 06-1469par/herg0

61

winNIMMAlTM,

Editorial Group for HighPerformance Computing andCommunications 1995

Editor

Sally E. HoweNational Coonlination Office

Writing Group

Norman S. Glick. NSAStephen Griffin. NSFFrederick C. Johnson. NISTThomi,s A. Kitchens. DOEDaniel R. Masys. NIHJoan H. Novak, EPAAlexis T. Poliakoff, EDThomas N. Pyke, Jr.. NOAAPaul H. Smith,. NASAStephen L. Squires. ARPAPatricia N. Williams. NCO

Copy Editor

Patricia N. WilliamsNational Coordination Office

Acknowledgments

Many people contributed to this hook, and wethank them for their efforts. We especially thankJoe Fitzgerald and Troy Hill of the AudiovisualProgl am Development Branch at the NationalLibrary of Medicine for their artistic contribu-tions and for the preparation of the hook in itsfinal form, aild we thank Patricia Carson andShannon Uncangco of the National CoordinationOffice for their contributions throughout thepreparation of this hook.

62