By Timothy Gardner and Max A. Cherney
WASHINGTON/SAN FRANCISCO (Reuters) -The U.S. has formed a $1 billion partnership with Advanced Micro Devices to construct two supercomputers that will tackle large scientific problems ranging from nuclear power to cancer treatments to national security, Energy Secretary Chris Wright and AMD CEO Lisa Su told Reuters.
The U.S. is building the two machines to ensure the country has enough supercomputers to run increasingly complex experiments that require harnessing enormous amounts of data-crunching capability. The machines can accelerate the process of making scientific discoveries in areas the U.S. is focused on.
Energy Secretary Wright said the systems would “supercharge” advances in nuclear power and fusion energy, technologies for defense and national security, and the development of drugs. Scientists and companies are trying to replicate fusion, the reaction that fuels the sun, by jamming light atoms in a plasma gas under intense heat and pressure to release massive amounts of energy. “We’ve made great progress, but plasmas are unstable, and we need to recreate the center of the sun on Earth,” Wright told Reuters.
“We’re going to get just massively faster progress using the computation from these AI systems that I believe will have practical pathways to harness fusion energy in the next two or three years.” Wright said the supercomputers would also help manage the U.S. arsenal of nuclear weapons and accelerate drug discovery by simulating ways to treat cancer down to the molecular level. “My hope is in the next five or eight years, we will turn most cancers, many of which today are ultimate death sentences, into manageable conditions,” Wright said.
The plans call for the first computer called Lux to be constructed and come online within the next six months. It will be based around AMD’s MI355X artificial intelligence chips, and the design will also include central processors (CPUs) and networking chips made by AMD. The system is co-developed by AMD, Hewlett Packard Enterprise, Oracle Cloud Infrastructure and Oak Ridge National Laboratory (ORNL).
AMD’s Su said the Lux deployment was the fastest deployment of this size of computer that she has seen.
“This is the speed and agility that we wanted to (do) this for the U.S. AI efforts,” Su said.
ORNL Director Stephen Streiffer said the Lux supercomputer will deliver about three times the AI capacity of current supercomputers.
The second, more advanced computer called Discovery will be based around AMD’s MI430 series of AI chips that are tuned for high-performance computing. This system will be designed by ORNL, HPE and AMD. Discovery is expected to be delivered in 2028 and be ready for operations in 2029.
Streiffer said he expected enormous gains but couldn’t predict how much greater computational capability it would have.
The MI430 is a special variant of its MI400 series that combines important features of traditional supercomputing chips along with the features to run AI applications, Su said.
The Department of Energy will host the computers, the companies will provide the machines and capital spending, and both sides will share the computing power, a DOE official said.
The two supercomputers based on AMD chips are intended to be the first of many of these types of partnerships with private industry and DOE labs across the country, the official said.
(Reporting by Max A. Cherney in San Francisco; Editing by Tom Hogue)
Comments