Approach

The sponsoring agencies, the organizing committee, and the program committee organized the workshop to emphasize technical deliberations among a selected group of experts to consider the technical ramifications of the challenging goals set before them. An environment was created that permitted intense focus time of small groups of individuals wrestling with complicated issues, conflicting goals, and severe uncertainties. But, frequent opportunities for interdisciplinary discussions were arranged to enable the richest possible context for exploring the myriad issues.

Four working groups were established with chair and co-chairs determined prior to the meeting. These were the

• Applications and Algorithms Working Group
  • Chair - Geoffrey Fox
  • Co-Chair - Rick Stevens
• Device Technology Working Group
  • Chair - Carl Kukkonen
  • Co-Chair - John Neff
  • Co-Chair - Joe Brewer
  • Co-Chair - Doc Bedard
• Architecture and Systems Working Group
  • Chair - Harold Stone
  • Co-Chair - Thomas Sterling
• Software Technology Working Group
  • Chair - Bert Halstead
  • Co-Chair - Bill Carlson

Each of the working groups was comprised of approximately a dozen experts from industry, academia, and government representing experience in the implementation, applications, and research of high-performance computing systems, and their future requirements as viewed in the context of the Federal agency missions.

Each working group was directed by the program committee to consider a number of questions as they related to the specific topic areas. While not every question was exactly germane to all the working group subjects and it was often difficult or impossible to give precise or complete responses in light of current knowledge, these questions had a strong and positive influence on focusing the participants on issues that led to the important findings. The questions addressed were

1. What is the state of the art in your area?
2. What is the level of investment being made ($, work force, facilities)?
3. What is the current focus of advanced work?
4. Define key metrics of figures of merit that measure scale/capability in your area. In such terms, estimate the current rate of progress.
5. What are the barriers and challenges?
6. List the potential opportunities for significant advances that may overcome these major barriers and challenges. Include both evolutionary and revolutionary advances. Indicate primary intermediate accomplishments that would exemplify progress in those directions.
7. In chart form, please relate figures of merit to intermediate accomplishments and estimate the most likely rate of progress against calendar years through the end of the decade.
8. What order-of-magnitude investment in funding, time, and human resources is required to achieve goals assuming favorable outcome of experimental work?

The workshop met in a series of plenary and working sessions. The former (five) questions were intended to force sharing of evolving thinking across working groups. The latter were to provide the concentrated time for deep examination issues of the independent groups. Often, these split into splinter groups to work separately on specific core or critical path issues.

Prior to the workshop each working group formulated its respective position on the state of the art in its field and the key problems as currently perceived. At the first plenary session, the chairs of each working group presented these position statements to the entire workshop to establish a shared context. Also, at the inaugural meeting Seymour Cray and Konstantin Likharev each gave a presentation to set the tone of the workshop and to initiate the debate that was to continue for the remaining three days.

At the conclusion of the workshop, each working group presented its closing positions and identified questions left unresolved. Contributors worked together to write the final report for each working group. These reports are included in Part II of this report. Also, the program committee worked to identify cross-cut issues and findings and to synthesize them into a single coherent structure. These are presented in Part III of the report.