Hewlett Packard Enterprise (HPE) has delivered the fastest supercomputer, El Capitan, to the Lawrence Livermore National Laboratory (LLNL), marking HPE’s distinction as building only three exascale systems in the world. El Capitan, running at 1.742 exaflops and achieving 58.89 gigaflops performance per watt, is also one of the top 20 most energy-efficient systems in the world. It will allow the United States to maintain a competitive edge in national security and enable the National Nuclear Security Administration’s Tri-Labs – LLNL, Sandia National Laboratories, and Los Alamos National Laboratory – to meet the increasingly demanding requirements for ensuring the safety, security, and reliability of the nation’s nuclear stockpile. LLNL intends to leverage artificial intelligence (AI) models for classified and unclassified workloads.
El Capitan will address secondary missions in maintaining national security, such as nuclear nonproliferation and counterterrorism, and be used for material discovery, nuclear data, and high energy density science, such as inertial confinement fusion research conducted at the National Ignition Facility. Insights gained from research performed on El Capitan will also feed into unclassified projects in the realm of energy security, climate change, power grid modernisation, drug discovery, and other areas.
The innovative platform enabling El Capitan’s performance is based on advanced technology anchored by HPE’s leadership-class supercomputing solution, HPE Cray EX equipped with the AMD Instinct™ MI300A APUs that integrate CPU cores and GPU cores with high-bandwidth memory into one package, the HPE Slingshot interconnect, and a custom storage solution. HPE Slingshot, an ethernet-based high-speed fabric, serves as the backbone to El Capitan’s collective power, enabling large calculations to be performed across the system’s more than 11,000 nodes. As part of the public-private partnership, LLNL and HPE co-developed a custom near-node local storage solution to reduce latency that is dynamically configurable and tied to a global Lustre-based file system shared across all compute nodes.
El Capitan was designed to prioritise energy efficiency and leverages a high-density design made possible by HPE’s industry-first 100 per cent fanless direct liquid cooling system architecture. As the world leader in supercomputing, HPE delivers an optimised end-to-end solution inclusive of software, storage, and services that helps customers advance scientific innovation and discovery in the AI era.
