US Dept of Energy partners with AMD to build two supercomputers: Report

To build two supercomputers that will address pressing scientific issues like nuclear power, cancer treatments, and national security, the United States has established a $1 billion partnership with Advanced Micro Devices (AMD).

Energy Secretary Chris Wright and AMD CEO Lisa Su were cited for the new partnership when the Reuters news agency first reported.

Recommended Stories

list of 4 itemsend of list

The US is building the two supercomputers to ensure that there are enough to conduct increasingly complex experiments that call for the use of enormous amounts of data-crunching power. The advancement of scientific discoveries in the US’s research is made more quickly thanks to the machines.

The developments in nuclear energy, fusion energy, defense and national security, and drug development were “supercharged,” according to Energy Secretary Wright.

By releasing a lot of energy by jamming light atoms in a plasma gas under intense heat and pressure, scientists and businesses are trying to replicate fusion, the process that causes the sun. We’ve made great strides, but the stability of plasmas requires a resemblance to the sun’s core, according to Wright.

“We’re going to make just a lot faster progress using these AI systems’ computation,” according to the statement from the company. “I think these AI systems will have practical ways to harness fusion energy in the next two or three years.”

By simulating treatments for cancer at the molecular level, Wright claimed that the supercomputers would also aid in managing the US arsenal of nuclear weapons and aid in drug discovery.

“It’s my hope that most cancers, many of which are currently the most serious ones, will be manageable,” Wright said.

“Stimility and speed”

Within the next six months, the first computer, called Lux, will be built and operational, according to the plans. It will be based on the AMD MI355X artificial intelligence chips, as well as AMD’s MI355X networking chips and central processing units (CPUs). Hewlett-Packard Enterprise (HP), Oracle Cloud Infrastructure, and Oak Ridge National Laboratory (ORNL) collaborate on the system.

Su of AMD claimed that this computer’s Lux deployment was the fastest she had seen.

Su remarked, “This is the speed and agility we desired for the US AI efforts.”

The Lux supercomputer, according to ORNL Director Stephen Streiffer, will have an AI capacity that rivals that of the current supercomputers.

The second, more sophisticated computer, called Discovery, will be based on AMD’s MI430 series of AI chips, which are designed for high-performance computing. ORNL, HP, and AMD will design this system. The discovery is anticipated to be finished in 2028 and operational by 2029.

Streiffer predicted significant improvements, but he was unable to say how much more computational power it would have.

According to Su, the MI430 is a unique variant of AMD’s MI400 series that combines the features required for AI applications with important characteristics of traditional supercomputing chips.

According to a DOE official, both parties will share the computing power and the companies will provide the computers and the US Department of Energy (DOE) will host the computers.

The two supercomputers, which are based on AMD chips, are intended to be the first of many of these collaborations, according to the official.