El Capitan supercomputer is ready to handle nuclear stockpile and AI workflows
Garry McLeod/LLNL
The fastest supercomputer in the world, housed at Lawrence Livermore National Lab, was dedicated to U.S. national security missions on Thursday, supported by novel hybrid processing chips.
LIVERMORE, Calif. — The El Capitan supercomputer at Lawrence Livermore National Laboratory was officially dedicated to U.S. national security missions and nuclear stockpile management Thursday.
During El Capitan’s dedication ceremony at Lawrence Livermore National Laboratory in California, stakeholders in both the public and private sector highlighted El Capitan’s distinction as the world’s fastest supercomputer, boasting the ability to compute over 2.79 exaflops of data. The ceremony itself reiterated the $600 million dollar machine’s focus on stewarding management of the U.S. nuclear stockpile and other national security research areas, along with broader scientific topics.
“Computing has really been a part of our laboratory DNA since it was founded,” said LLNL Director Kimberly Budil. “This relationship between learning or experimentation and using computing to take that to the next level has been there since the beginning. We're very excited to see what we can do with this incredibly advanced technology.”
The secret to El Capitan’s speed lies within both its hardware and software. Hewlett Packard Enterprise supplied the exascale machinery and memory technology, and the software is provided by supercomputing chip manufacturer AMD.
The chip in particular brings a unique approach to data processing, as the accelerated processing unit combines the capabilities of a graphics processing unit with a standard central processing unit. Combining these capabilities results in El Capitan’s ability to offer high resolution images for scientific models and simulations, offering more precision into various research domains.
AMD CEO Lisa Su said that the partnership with Energy in creating El Capitan encouraged the company to further develop its APUs.
“It was a big bet. It was an important bet,” Su said. “We had a decision to make whether we wanted to really push a module for the next innovation.”
El Capitan will represent the first exascale computer dedicated to the mission of the National Nuclear Security Administration.
“It represents a pivotal next step in our commitment to ensuring the safety, security and reliability of our nation's nuclear stockpile without the need to resume underground nuclear testing,” NNSA Acting Deputy Administrator Corey Hinderstein said at a November briefing before the SC24 conference.
“This machine's power will enable us to incorporate various real-world factors, such as materials, manufacturing imperfections, environmental conditions and abnormal and hostile environments,” LLNL’s Weapon Simulation and Computing Associate Director Rob Neely said at the November briefing. “This means more accurate predictive capabilities and, by extension, better informed decision making for the NNSA stockpile stewardship program.”
Artificial intelligence will also be present in El Capitan’s future workloads, intersecting with other scientific and research domains. As other national labs work to incorporate AI into analyses and simulation capabilities, stakeholders anticipate that El Capitan will address similar computations, namely with materials sciences and physics problems, as well as how AI and large language models can function in the national security environment.
“You can imagine in the future, a relationship between big computers and experimental science where you use AI … to drive decisions about what the next experiment should be,” Budil said. “We're doing a huge amount of work on AI and machine learning right now, and in two years, I guarantee you, we'll be doing things we don't know about today.”
Pat Falcone, deputy director for Science and Technology at LLNL, said that innovation in generative AI in particular stands to further what supercomputing can do in research.
“Today, rapid advancement [in] generative artificial intelligence are opening new possibilities for us at the lab to deploy powerful computing capabilities into critical research spaces and develop fundamental efforts in the broader realm of artificial intelligence,” he said.
The completion of El Capitan at LLNL marks the latest installment of supercomputing within the U.S. national lab network, with fellow labs like Los Alamos unveiling large scale computers for federal research.