SC17 Denver, CO

Exascale Challenges and Opportunities

Authors: Dr. James Ang (Sandia National Laboratories)

Abstract: This BoF explores the challenges and opportunities of Exascale Computing. The motivations for exascale are centered on drivers such as economic competitiveness, national security, health care, energy, manufacturing, materials, and science. This session gathers a forum of HPC experts to identify what they consider to be key challenges in achieving exascale computing and how they would best propose to meet them, with special attention to collaboration opportunities. A few of the key topics include: exascale eco-system elements, lessons from establishing public/private partnerships, impact of exascale R&D investments on the broader HPC community. Significant audience participation will be encouraged.

Long Description: This Birds of a Feather session will provide a forum for HPC experts to discuss amongst themselves and with the SC17 audience, the key challenges and opportunities for achieving exascale computing. The following HPC experts are committed to participate in this BoF:

• Bill Gropp, NCSA Director and Professor of Computer Science, University of Illinois at Urbana Champaign;

• Satoshi Matsuoka, Division Lead of Supercomputing Infrastructure at GSIC, as well as the Laboratory Director of Joint AIST-Tokyo Institute of Technology, Open Innovation Laboratory on Real World Big Data Computing;

• Mateo Valero, Director Barcelona Supercomputing Center, and Professor Universitat Politécnica de Catalunya;

• Paul Messina, Director U.S. Exascale Computing Project, and Distinguished Fellow, Argonne National Laboratory.

This group of experts will set the stage to engage each other and the audience in the discussion of a number of key questions on exascale computing:

• Do you think it is possible to reach exascale with commodity technology processors? Why or why not?

• The convergence of HPC and Data Analytic Computing. Is this good or bad? Why?

• If trends in I/O performance continue – which lags computing performance increases, will exascale systems be productive? Why or why not?

• What is the role of architecture R&D, non-recurring engineering and leading-edge system acquisitions in achieving exascale? What is an appropriate balance of funding for these activities?

• What is the role of public/private partnerships in achieving exascale?

• Do exascale R&D investments have impact on the broader HPC user community? How?

Conference Presentation: pdf

Birds of a Feather Index