Commentary - The Exascale Evolution

The Exascale Evolution

Lawrence Livermore was founded in 1952 and almost immediately installed its first high-performance computing (HPC) system. The UNIVAC-I was a kiloscale computer, which means it could perform a thousand operations per second—a huge boost over human calculation speed. A later version of UNIVAC ran simulations of hydrodynamics for thermonuclear weapon design.

Our commitment to HPC leadership and investment spans Livermore’s 73-year history, and we’ve come a long way since the early vacuum-tube and punch-card machines. The Laboratory began working with IBM in the 1950s and other companies during the following decades to build an ever-expanding roster of world-leading supercomputers. Our first parallel computing architecture came online in the 1970s. The phrases massively parallel and distributed memory entered the lexicon a decade later. Just before this millennium began, we reached the terascale threshold, where computers performed a trillion operations per second.

Computing provides a uniquely versatile tool for understanding the world, and as our mission has evolved alongside the changing national security landscape, the need for exceptional HPC capabilities has only increased. It’s difficult to exaggerate the importance of high-performance simulations to stockpile stewardship, multi-domain deterrence, high-energy-density physics, advanced manufacturing and materials design, climate and energy security, biosecurity and bioresilience, and myriad other scientific pursuits.

This issue of Science & Technology Review introduces El Capitan, the next step in our journey of computing leadership. El Capitan is the Laboratory’s—and the National Nuclear Security Administration’s (NNSA)—first exascale supercomputer, capable of more than two quintillion calculations per second. A system of this scale requires years of planning, coordination, and preparation. El Capitan will play a central role in our stockpile stewardship mission by helping scientists delve deeper into the physical phenomena inherent in transport problems, radiation hydrodynamics, and fusion reactions.

In 2024, three of El Capitan’s early access systems debuted on the international Top500 list of most powerful supercomputers. Before the full exascale system was completed, we had 14 machines on the list—by far the most reported by any computing center in the world. Indeed, even our relatively “old” machines are still powerful enough to claim top spots on the list. Deployed in 2018, Sierra is number 12. El Capitan is, for now, number 1.

While worldwide rankings are gratifying, they fail to capture what’s truly special about this machine. El Capitan is remarkable not only for its technological innovations, but more importantly, for the scientific and engineering discoveries and predictions it will enable us to make. As the feature article on pp. X–X describes, we have improved our ability to deliver on the U.S. Department of Energy’s (DOE) and NNSA’s missions. We have upgraded our computing power by orders of magnitude, producing higher resolution simulations faster than ever before. We will now be able to run ensembles of highly detailed, three-dimensional simulations of extremely complex physics. This capability will shape our designs, inform our experiments, expose potential flaws, and add confidence to our high-consequence work. We’re upholding our responsibility to national security by integrating superior HPC capabilities into our experiments and scientific workflows and, in doing so, staying ahead of our adversaries. El Capitan is also an extraordinary machine for machine learning, which will allow us to develop entirely new scientific workflows.

Our journey to this milestone was long, complicated, and not always smooth—but always exciting. From facility upgrades to system operation, the feature article emphasizes the importance of collaboration in bringing this enormous project to fruition. It takes a village to design, procure, install, and shake out a system of this complexity and scale, and I’m continually impressed by the deep knowledge and excellence demonstrated by teams in Livermore Computing and their partners across the Laboratory.

We are living in an era of astonishing technological and scientific progress as exemplified by El Capitan as well as other activities at the Laboratory. This issue’s other articles highlight the diversity of Livermore’s mission-relevant efforts. One team has designed innovative reactors for ammonia synthesis, which is a key component in the global food supply chain. In the Environment, Safety, and Health Directorate, monitoring and protecting the Laboratory’s natural environment go hand in hand with sustainability practices. Thank you for joining me on a journey where, as this entire S&TR issue illustrates, the future happens now.