SC97: Panels and Town Hall Meetings

SC97 Panels and Town Hall Meetings

Participate in SC97! The Panels and Town Hall meetings are designed for attendee involvement. This is your opportunity to ask questions of experts in high performance networking and computing, or to make your voice heard about future SCing conferences. Education panels offer perspectives on the use of HPC in the classroom and beyond, for life-long learning.

Sunday	Monday	Tuesday
Wednesday	Thursday	Friday

Sunday, November 16
3:30-5pm		Education Panel: Collaborative Environments and Technology Participants: Beau Fly Jones, Ohio Supercomputer Center; Geoffrey Fox, NPAC, Syracuse University; David Emugh, NCSA


		This panel will address the issue of the development of collaborative learning environments, the associated technologies, and hidden problems in their use.
Monday, November 17
8:30-10am		Education Panel: Standards, Frameworks, and Technology Planning for Schools Participants: Ralph Annina, Senior Consultant, Allied Computer Group; Carol Doherty, Director, Professional Development, Center for Innovation in Urban Education


		This panel will address the issues surrounding the impact of national math, science, and technology education standards on technology planning for schools.

10:30am-noon		Education Panel: The Future Impact of the National Partnerships in Computational Infrastructure on Education Participants: Roscoe C. Giles, Boston University; John Ziebarth, NCSA; Kris Stewart, SDSU, NPACI


		This panel will focus on the next decade of HPC and its impact on education and society. HPC has had a significant impact in the past; we are now at a transition point developments to be seen over the next 10 years.

3:30-5pm		Education Panel: Development and Assessment of Virtual Reality Materials for Education Participants: Chris Dede, George Mason University; Robert Fixen, Montana State University; Melissa Kelly, NCSA


		This panel will evaluate the application of virtual reality technology in the classroom and other educational settings.
Tuesday, November 18
10:30am-noon		Panel: Digital Libraries Moderator: Sally Howe, National Coordination Office for Computing, Information, and Communications Panelists: Alexa McCray, National Library of Medicine; Terence Smith, University of California at Santa Barbara; Howard Wactlar, Carnegie Mellon University Room A1


		"Digital libraries" is a rapidly advancing field that will place heavy demands on high performance computing and communications technologies and infrastructure. The panelists will describe the demands emanating from the technologies they are developing to organize, find, and extract knowledge from digital information: Terry Smith is principal investigator of the Alexandria Project, one of six NSF/DARPA/NASA Digital Libraries Research Initiative projects. Alexandria allows users to locate maps and other spatially-index material in geographically dispersed digital libraries and databases. Howard Wactlar is principal investigator of the Informedia project, another of the six Initiatives. That project is integrating speech, image, and natural language understanding to create and explore digital video libraries. Alexa McCray leads NLM in developing advanced technologies for extracting knowledge from biomedical informatics resources including the Library's Unified Medical Language System, Internet Grateful Med, and Visible Human databases. The panelists will also discuss the potential impact of technologies developed in a possible Digital Libraries II initiative, whose proposed focus will be on developing tools that make the libraries easy to use by their highly diverse user communities and on content.

1:30-3pm		Panel: High Performance I/O Moderator: Jim Pool, Center for Advanced Computing Research Panelists: Paul Messina, Jim Pool, Center for Advanced Computing Research; David Greenberg, Sandia National Laboratories; Reagan Moore, NPACI; Marc Snir, IBM Thomas J. Watson Research Center; Tom Ruwart, University of Minnesota Room A1


		While transferring data among the components of a computing system has always been a challenge to system architects, continuing increases in processor speed now make high performance I/O essential to achieving potential system performance. Quantitative investigations of I/O-intensive scientific and engineering applications on large scalable parallel systems have demonstrated dramatic imbalances between compute power and I/O, in addition to mismatches between system design and application practice. Increased use of scalable parallel systems for data-intensive applications-both commercial and emerging scientific and engineering applications-introduces additional problems including network I/O and transfers to and from archival storage systems. Processor speeds have generally outpaced the performance of other system components, making sequential approaches to I/O inadequate and parallelism in I/O a necessity. Since I/O involves many hardware and software components, introducing parallel I/O and avoiding possible bottlenecks is a difficult design problem. Moreover, effective use of parallel I/O currently imposes a burden on the application developer since minor changes in parameters or options can have major impacts on I/O performance. Panelists representing large-scale applications, system vendors, and computer science research projects will review requirements for high performance I/O and speculate on possible approaches to pragmatically satisfying these requirements.

1:30-3pm		Education Panel:Presentation of the UCES Awards Moderator: Jim Corones, AmesLab


		Each year, the Department of Energy has sponsored an awards program for excellence in undergraduate computational engineering and science. During this panel each of the 1997 UCES awardees will make a short presentation on his/her work and receive the UCES award. http://uces.ameslab.gov/uces/awards

3:30-5pm		Panel: Building Computational Grids Moderator: Ian Foster, Argonne National Laboratory Panelists: Carl Kesselman, USC Information Sciences Institute; Larry Smarr, NCSA; Songnian Zhou, Platform Computing Corporation Room A1


		Developments in networking and software technologies are enabling the integration of geographically distributed compute, storage, network, and other resources into high performance distributed computing systems or computational grids. These computational grids promise to revolutionize high performance computing by enabling both the construction of new tools providing qualitatively new capabilities, and wider access to unique resources such as supercomputers, scientific databases, and instruments. This panel will discuss a variety of issues relating to the construction of computational grids, including: How can computational grids be expected to change science and engineering? What will they look like? What are the major obstacles to creating and using grids? Where are research advances must urgently required? Can we expect the required advances to come from the commercial sector, or do research problems remain? The panelists include Carl Kesselman, co-leader of the Globus research project, which is investigating advanced grid technologies; Larry Smarr, director of the National Computational Science Alliance and a leading advocate of the grid concept; and Songnian Zhou, chief technology officer at Platform Computing Corporation, which develops grid software.

3:30-5pm		Education Panel: Internet in the K-12 Classroom: The Realities of Technology Transfer Participants: TBA


		This panel will deal with specific issues concerning the integration of technology into the curriculum and the classroom.

5-6:30pm		Town Hall Meeting with the Presidential Advisory Committee on HPCC, IT and NGI Moderator: Ken Kennedy, CRPC, Rice University Room A1


		In February, President Clinton established the Presidential Advisory Committee on HPCC, IT and NGI, headed by Ken Kennedy of the Center for Research on Parallel Computation and Bill Joy, co-founder and current vice-president for research at Sun Microsystems. The committee provides guidance and advice on all areas of high performance computing, communications, and information technologies as they relate to the administration's efforts to accelerate development and adoption of information technologies that will be vital for American prosperity in the 21st century. In this Town Hall meeting, SC97 participants will have an opportunity to ask questions of several representatives of this committee. Attendees' questions submitted prior to the Town Hall meeting will be answered first, followed by an open mike session for questions from the floor. The committee will also be looking for input on their areas of focus and future actions.
Wednesday, November 19
10:30am-noon		Panel: Confidence in Simulation Moderator: Vic Reis, Department of Energy Panelists: Tom Adams, Los Alamos National Laboratory; Richard Hodur, Naval Research Laboratory; Greg Shubin, The Boeing Company; Sam Thurman, Jet Propulsion Laboratory Room A1


		This panel will explore the challenges, new opportunities, and limitations of high-fidelity numeric simulation. The panelists will address how simulation results are compared with actual physical phenomena and how results are projected when physical tests are not possible. With the variables imposed by the accuracy of the model, the accuracy of the measurements, and the fundamental physical limitations, how well can we do--and when do we decide that the results are good enough? The software validation for the entry, descent, and landing for the Mars Pathfinder, along with simulations related to ASCI, numerical weather prediction, and aircraft design and manufacture will be discussed.

1:30-3pm		Panel: Breakthroughs and Challenges ahead in Computer Architecture Moderator: David Culler, University of California, Berkeley Panelists: James H. Gray, Microsoft; Greg Papadopoulos, Sun Microsystems Computer Company; Burton Smith, Tera Computer; John L. Hennessy, Stanford University Room A1


		In recent years we have seen a technological turnover in high performance computer architecture, followed by increasing levels of commoditization and consolidation in the industry. Meanwhile, we have approached the brink of new technological epochs, such as processor and DRAM on chip, multi-gigabit fiber links, ten thousand processor systems, lithography limits, and practical speed of light constraints. This panel will explore views on where computer architecture is headed, what breakthroughs may lie around the corner, and what are likely to be the critical challenges ahead.

3:30-5pm		Panel: The Future of Software Development Environments for Parallel/Distributed Computing Moderator: Barton Miller, University of Wisconsin Panelists: Dennis Gannon, Indiana University; Ken Kennedy, Rice University; Doug Pase, IBM Power Parallel Division; Dan Reed, University of Illinois Room A1


		The parallel and distributed community has suffered from the lack of powerful, ubiquitous programming tools. While there have been some notable successes, on the whole, most programmers find limited choices on most platforms. The question affects compilers, libraries, profilers, debuggers, and visualizers. The panel will present their views on this issue, addressing questions that include: What is your vision for the route to ubiquitous quality tools? What platforms should we be targeting? What is the future of UNIX vs. Windows/NT vs. custom systems? Is portability important or even possible? Is there a best language or programming model? Is there a commodity programming language or programming model? The panelists include Dennis Gannon, leader of the HPC++/SAGE++ project; Ken Kennedy, director of the Center for Research on Parallel Computation (CRPC); Doug Pase, Tools Architect for IBM RS/6000 Power Parallel Division; and Dan Reed, leader of the Pablo project.

5-6:30pm		Town Hall Meeting on the Internet: Then, Now, and Tomorrow Moderators: Dan Lynch, Founder of Interop; Katie Hafner, Newsweek, Silicon Valley Correspondent Invited Participants: Irving Wladawsky-Berger, IBM; Vint Cerf, MCI; Craig Partridge, GTE/BBN; Don Nielson, SRI International; Jordan Becker, AOL/ANS; Steve Wolff, Cisco; Al Weis, Advanced Network Services; Brian Reid, Digital Equipment Corporation; Bill Joy, Sun Microsystems; Bob Kahn, Corporation for National Research Initiatives; Larry Roberts, ATM Systems; Paul Mockapetris, Software.Com; George Strawn, National Science Foundation; Len Kleinrock, University of California, Los Angeles; Larry Smarr, NCSA; Van Jacobson, Lawrence Berkeley National Laboratory; Dave Clark, Massachusetts Institute of Technology; Larry Landweber, University of Wisconsin; Dave Farber, University of Pennsylvania Room A1


		Spend an hour with any or all of the founders of the Internet. A brief presentation by four of the originals will be followed by a free-ranging open microphone session for questions, reminiscences, and dreams of the future.
Thursday, November 20
10:30am-noon		Panel: Networking Initiatives: Internet 2 and Next Generation Internet Moderator: Larry Landweber, University of Wisconsin Panelists: David J. Farber, University of Pennsylvania; George Strawn, National Science Foundation; Jim Gray, Microsoft; Steve Wolff, Cisco Systems Room A1


		Internet 2 (I2), a project of more than 100 U.S. research universities, and Next Generation Internet (NGI), a multi-agency program of the U.S. government, are separate but related initiatives. Their goals include the development of new network-based applications and the prototyping of networks that incorporate new pre-competitive network technologies to support the different service levels required by these applications. NGI also includes a research component that will look further into the future. In addition, to support the various components, both NGI and I2 have as a goal the connection of U.S. research organizations to backbones at significantly higher data rates than is now the case.

10:30am-noon		Town Hall Meeting on SC Conferences Moderator: Cherri Pancake, Oregon State University


		On Thursday, November 20, from 5-6:30pm, attendees will have the chance to interact with committee members from SC97, SC98, SC99, and SCXY (the SC conference series' steering committee), discuss this year's conference, and make suggestions for future years' conferences. This is also the ideal opportunity for attendees to express their interest in serving on one of the conference's diverse committees. Plan to attend and make your voice heard.

3:30-5pm		Panel: Visualization Moderator: Paul Woodward, University of Minnesota Panelists: Pat Hanrahan, Stanford University; Paul Woodward, University of Minnesota; Chris Johnson, University of Utah; Tom DeFanti, University of Illinois at Chicago; Philip Heermann, Sandia National Labs Room A1


		DoE's ASCI program, NSF's PACI program, the DoD HPC modernization program, and NASA's grand challenge program are accelerating the pace of supercomputer hardware development. As a result, researchers are running much larger simulations and are therefore generating much larger data sets for visualization. This panel will discuss the present bottlenecks that are limiting the rate at which visualization systems and software can accelerate in order to keep pace with the new, faster growth in demand for data visualization.
Friday, November 21
8:30-10am		Panel: PACI Moderator: Bob Borchers, National Science Foundation Panelists: Sid Karin, National Partnership for Advanced Computational Infrastructure; Peter Kollman, University of California, San Francisco; Larry Smarr, National Computational Science Alliance; Bob Sugar, University of California, Santa Barbara Room A1


		In 1985 the National Science Foundation (NSF) established a group of supercomputing centers to further research using computational methods. From 1985 to 1997, partly through the contributions of these centers, the US saw computation become a widely used tool in many disciplines, an explosive growth of the Internet, the birth of the Web, and tremendous increases in computing power. In 1995, in response to the changing technological landscape, the NSF initiated a competition to establish Partnerships for Advanced Computational Infrastructure (PACI). With PACI, starting in October of 1997, the focus will shift from centralized facilities to nationwide partnerships, reflecting the growing importance to computational science of computer science, high-speed networking, distributed computing, and remote collaboration. Two partnerships, the National Computational Science Alliance led by the University of Illinois at Urbana-Champaign and the National Partnership for Advanced Computational Infrastructure led by the University of California, San Diego, have joined dozens of research organizations across the country in large-scale efforts to improve the technological environment available to computational scientists. This panel will outline the current status of the PACI program, its resources, and future plans from both the perspectives of the principal investigators and those of scientists whose research depends on the use of computing resources.

8:30-10am		Panel: I/O and Interconnects Moderator: Bill Boas, Abba Technologies Panelists: Greg Chesson, Silicon Graphics; Dave Follett, Giganet; Randy Rettberg, Sun Microsystems; Justin Rattner, Intel; Chuck Seitz, Myricom Room A3


		This panel will be a survey of the current technologies and architectures in use or in development today for both internal SMP interconnects as well as for clustering multiple processors in shared and non-shared memory architectures. The discussion will focus on design trade-offs, limiting bandwidth factors, latency, applications and operating system interfaces, driver structures, media technologies, parallel vs. serial transmission, ASIC densities and limitations, speed, and number of gates. Technologies discussed will include Craylink, HiPPI-6400, SCI, Gigaplane, Myrinet, ATM, PCI evolution, and the emerging Virtual Interface Architecture.

10:30am-noon		Panel: ASCI Strategic Alliances Moderator: Alex Larzelere, Department of Energy Panelists: Michael T. Heath, University of Illinois;Daniel Meiron, California Institute of Technology;David Pershing, University of Utah; William Reynolds, Stanford University;Robert Rosner, University of Chicago Room A1


		In 1995 the U.S. Department of Energy (DOE) established the Accelerated Strategic Computing Initiative (ASCI). Its purpose is to bring together the resources of three DOE Defense Program National Laboratories (Los Alamos, Lawrence Livermore, and Sandia National Laboratories) in a consolidated effort to achieve significant advances in computational modeling and simulation capabilities available to meet national goals for nuclear stockpile stewardship. One of the five key ASCI strategies is to "Encourage Strategic Alliances and Collaborations." The ASCI Academic Alliances strategy was established to enhance the overall ASCI goals by establishing technical interactions between the ASCI Laboratories and leading-edge academic R&D in the United States. These alliances will take the form of Centers of Excellence working on key multi-disciplinary applications and high performance computing requiring major advances in large-scale computer-based simulation approaches. This panel will discuss the major goals and objectives of the Academic Alliances program, including the establishment and validation of the scientific methods of large-scale modeling and using simulation and computation as a cornerstone of the scientific methodology in SBSS-related applications requiring coupled complex simulation sequences.

10:30am-noon		Panel: The Role of High Performance Computing and Informatics in Pharmaceutical Research and Development Moderator: Frederick H. Hausheer, BioNumerik Pharmaceuticals, Inc. Panelists: Frederick H. Hausheer, BioNumerik Pharmaceuticals, Inc.; Babu Venkataraghavan, American Cyanamid Company; Eugene Fluder, Merck & Co. Room A3


		This panel will discuss current applications of computing and informational technologies being applied in the pharmaceutical industry to help discover and develop new drugs in less time. It is clear that new therapeutic targets will be discovered by the identification of genomic sequencing. The panel will discuss how high performance computing and information technologies may play a role in the translation of potential new targets into new drugs. The panelists are recognized industry experts with extensive experience in pharmaceutical discovery and development and the application of high performance computing and information technologies. Frederick H. Hausheer, is the CEO of BioNumerik Pharmaceuticals, Inc.; Babu Venkataraghavan, is director of Biomedical Research Computing at American Cyanamid Company, Medical Research Division, Lederle Laboratories; Eugene Fluder is senior investigator, Molecular Design and Diversity, at Merck.