University of Iowa Vice President for Research and Economic Development Daniel A. Reed is playing a key role in shaping the United States government’s plans for developing and deploying exascale technology by 2023.
Reed chairs a subcommittee of the U.S. Department of Energy’s (DOE) Advanced Scientific Computing Advising Committee (ASCAC). The group was charged with reviewing an exascale technology plan drafted by DOE and National Nuclear Security Administration (NNSA).
ASCAC, of which Reed is also a member, has been tasked by DOE and NNSA to review the “conceptual design for the Exascale Computing Initiative” and to deliver a report by September.
Exascale refers to computing systems capable of at least one exaFLOPS, or a billion billion calculations per second.
To reach that goal would require a hundred-fold increase in sustained performance over current computing capabilities, but the payoff would be enormous. Achieving that rate and processing power and sophistication, and developing the necessary infrastructure to back it up, could lead to enormous advancements in basic research, engineering, earth science, biology materials science, energy issues, and national security.
The subcommittee report was issued a day before President Barack Obama signed an Executive Order establishing a National Strategic Computing Initiative on July 29. The initiative seeks to create a cohesive, multi-agency strategic vision and federal investment strategy in high-performance computing.
Reed’s subcommittee gave the DOE/NNSA plan strong support, along with seven recommendations for strengthening management of the Exascale Computing Initiative (ECI). Details of the subcommittee’s report may be found at http://www.hpcwire.com/2015/07/28/doe-exascale-plan-gets-support-with-caveats/.
Reed said the ECI was well-crafted and noted DOE’s demonstrated ability to manage complicated, multi-stakeholder projects.
“Like any ambitious undertaking, DOE’s proposed exascale computing initiative involves some risks,” Reed said. “Despite the risks, the benefits of the initiative to scientific discovery, national security and U.S. economic competitiveness are clear and compelling.”
Among the challenges presented by exascale technology are energy efficiency, scalable system software, memory technology, and the need to develop intuitive software engineering tools for scientists that allow them to take full advantage of exascale systems.
The subcommittee draft report was approved with a final full ASCAC version expected in August. Slides used in the subcommittees presentation in Washington, D.C. may be viewed at http://science.energy.gov/~/media/ascr/ascac/pdf/meetings/20150727/Exascale_Computing_Initiative_Review.pdf
In addition to his role as vice president at the UI, Reed is University Computational Science and Bioinformatics Chair and a professor of computer science and electrical and computer engineering.
He is also a member of the National Research Council study on the future of cyberinfrastructure for the National Science Foundation, a complement to the DOE study. And he previously served as a member of the DOE Secretary of Energy Advisory Committee’s ad hoc task force to assess exascale computing strategies.
Serving with Reed on the ASCAC sub-committee are members Martin Berzins, University of Utah; Bob Lucas, Livermore Software Technology Corporation; Satoshi Matsuoka, Tokyo Institute of Technology; Rob Pennington, University of Illinois, retired; Vivek Sarkar, Rice University; and Valerie Taylor, Texas A&M University.