The Department of Energy (DoE) has four of the top ten fastest supercomputers in the world, and the agency’s investment in supercomputing is crucial to fighting climate change and bolstering the electric grid, according to DoE Chief Information Officer (CIO) Ann Dunkin.
At the General Services Administration’s (GSA) High-Performance Computing (HPC) Summit on April 20, Dunkin explained how HPC – which include the most powerful and largest scale computing systems – enables researchers to study complex systems in ways that would otherwise be impossible.
“We call DoE, not very modestly, the engine of innovation for the U.S. government and for much of the world,” Dunkin said. “Supercomputing is important to us because it allows us to achieve advances in a variety of fields that simply would no longer be possible without supercomputing.”
DoE is now at the point of exascale, Dunkin said, confirming that the Frontier Supercomputer at Oak Ridge National Laboratory (ORNL) is the fastest supercomputer in the world and is also the nation’s first exascale computer – able to compute over one quintillion floating point operations per second.
And yes, “that’s a one with 18 zeros for those of you who haven’t comprehended that,” Dunkin said.
Following Frontier on the June 2022 Top500 list of the world’s most powerful supercomputers: Summit, also at ORNL, ranks at number four; Sierra at Lawrence Livermore National Laboratory comes in number five; and Perlmutter at Lawrence Berkeley National Laboratory ranks as number seven.
Dunkin shared two exciting examples of modeling that DoE uses these supercomputers for. The first, she said, is climate modeling.
“The complexity of climate models continues to grow,” Dunkin explained. “If you look at the modeling that we do at DoE, the modeling that EPA [Environmental Protection Agency] does, that NOAA [National Oceanic and Atmospheric Administration] does, that others do, you know, there’s this incredibly rich set of data that encompasses decades of data collection – in some cases, even more than that.”
“That level of data feeds some really amazing simulation models that are able to help us understand both short-term actions in the climate and long-term actions in the climate,” she continued. “So, those are incredibly important for us to be able to look out at the future and understand how we’re going to behave in the future.”
Similarly, Dunkin said DoE has developed models of the electrical grid. She referenced the cascading failure of the electric grid in the early 2000s, as well as the recent 2021 failure of the grid in Texas.
“These things are not unknown to us and the models we built help us to better understand the grid, to manage the grid, and to help people respond in situations where we have the potential for cascading failures,” Dunkin said. “Modeling is a topic that’s going to help us tremendously.”