The Department of Defense has made HPC news twice in the last few days – in one, the Army will spend $32 million on supercomputing technology from composable infrastructure vendor Liqid; in the other, DOD will partner with the Department of Energy and Microsoft to develop AI algorithms to support natural disaster first responders.
In the deal with Colorado-based Liqid, the Army has purchased two supercomputers to improve data analytics for organizations “across the military branches requesting such services,” according to a story by Andrew Eversen in the online military publication C4ISRNET.
The computers will be located at the Army Research Laboratory’s DOD Supercomputing Resource Center at Aberdeen Proving Ground in Maryland, according DOD announcement earlier this month. The High Performance Computing Modernization Program, or HPCMP, provides advanced computing capabilities to the DOD’s R&D community.
“The primary customer base for the (HPCMP) has been the physics-based modeling, which are your weapons designers and that kind of stuff — it’s all based on physics effects,” Matt Goss, director of the center, told C4ISRNET. “Now it’s more folks are focused on ‘I got this data from this model, how can I interrogate it? The way I’ve done it in the past with Excel and 20-year-old script isn’t efficient anymore because I have so much of it.’”
The new hardware will add 15 petaFLOPs to the HPCMP’s supercomputing capability, bring its total HPC power to 97 petaFLOPS.
“These machines have the capability to do the physics-based modeling, but also have a significant capability for artificial intelligence and machine learning applications,” said Thomas Kendall, supercomputing acquisition lead with the HPCMP and senior technical adviser to the director of the Army Research Lab’s Computational and Information Sciences Directorate.
The Liqid systems are scheduled to be installed by mid-fiscal 2021, according to an HPCMP news release.
In addition, DOD has formed a partnership with DOE and Microsoft with the goal of providing “near-real-time data to improve the decision-making of first responders engaged in natural disasters and humanitarian assistance efforts,” according to a U.S. Army blog by David Vergun.
The blog quotes Michael J. Kratsios, DOD’s CTO, saying that five consortia have been organized to use “the best AI technology talent from industry to respond to humanitarian assistance and to mitigate natural disasters by protecting property and lives….”
Cheryl Ingstad, director of DOE’s AI and technology office, told Vergun an important aspect of the project “is understanding first responders’ needs and developing AI in such a way that they can easily understand and use it.”
Microsoft Chief Federal Technology Officer Susie Adams said the solutions coming out of the partnership will be shared with global partners.
“AI and machine learning enable solutions never thought possible,” she told Vergun, adding that the company now has approximately 1000 researchers working in AI R&D. “AI is a tool to augment human intelligence, not replace it. It’s about getting data to the right people at the right time as quickly as possible so they can make better-informed decisions.”
