yoshitaka272 - Fotolia
The United States stopped conducting nuclear weapons tests in 1992, instead turning to supercomputers to test weapons virtually. But the computing capability limited the government's work to 2D simulations -- until now.
Today, the U.S. announced plans to acquire an exascale computer in late 2022 that will be capable of 3D simulations, which will provide nuanced simulations needed to maintain and modernize the current arsenal of the country's nuclear weapons.
The U.S. Department of Energy has contracted with Cray Inc. for a 1.5 exaflops system valued at about $600 million. El Capitan will be the third DOE exascale computer system that uses Cray's architecture and interconnects.
Cray is being acquired by HPE for $1.3 billion, in a deal announced in May. This U.S. supercomputing contract helps to explain why HPE is buying the firm.
Cray says it now has agreements worth $1.5 billion for supercomputers built with its Shasta architecture, which enables different processor types and interconnects to work together. It also supports the convergence of different types of workloads, such as AI, simulations and modeling.
El Capitan, which will be used by the National Nuclear Security Administration, will be installed at Lawrence Livermore National Laboratory in Livermore, Calif.
Because the system can use a variety of CPUs and accelerators, the machine will be tailored to fit the DOE's needs. A decision as to what processors will be included in its El Capitan system will be finalized later, said Peter Ungaro, Cray's president and CEO.
Why the U.S. needs exascale computers
The DOE has responsibility for the nation's nuclear stockpile. That responsibility, which includes maintaining and modernizing nuclear weapons, has made it a major buyer of the largest and most advanced supercomputers.
Bill GoldsteinDirector, Lawrence Livermore National Laboratory
"Ever since we ceased nuclear testing in 1992, it was clear that the nation would require massive increases in computing power," said Bill Goldstein, director of the Lawrence Livermore National Laboratory. Supercomputing systems enable the DOE to test and modernize weapons virtually.
The U.S. is focused on improving its existing nuclear weapons, unlike Russia and China.
"The Russians are fielding brand new nuclear weapons and bombs," said Lisa Gordon-Hagerty, undersecretary for nuclear security at the DOE. She said "a very large portion of their military is focused on their nuclear weapons complex."
It's the same for China, which is building new nuclear weapons, Gordon-Hagerty said, "as opposed to the United States, where we are not fielding or designing new nuclear weapons. We are actually extending the life of our current nuclear weapons systems." She made the remarks yesterday in a webcast press conference.
A thousandfold increase in compute power
An exascale computer provides a thousandfold increase in computing power from petascale, which was reached in 2008. Exascale systems are the leading edge in supercomputing and the technology will ultimately find its way into lower cost business systems.
El Capitan's 1.5 exaflops means it will be capable of 1.5 quintillion calculations per second. The system will use about 30 megawatts of power, which is within the range of power usage targeted for these systems.
Businesses use 3D simulation to design and test new products in high performance computing. That is not a unique capability. But nuclear weapon development, particularly when it involves maintaining older weapons, is extraordinarily complex, Goldstein said.
The DOE is redesigning both the warhead and nuclear delivery system, which requires researchers to simulate the interaction between the physics of the nuclear system and the engineering features of the delivery system, Goldstein said. He characterized the interaction as a new kind of problem for researchers and said 2D development doesn't go far enough. "We simply can't rely on two-dimensional simulations -- 3D is required," he said.
Nuclear weapons require investigation of physics and chemistry problems in a multidimensional space, Goldstein said. The work is a very complex statistical problem, and Cray's El Capitan system, which can couple this computation with machine learning, is ideally suited for it, he said.