At the SC24 conference, Hewlett Packard Enterprise (HPE) announced the delivery of El Capitan, the world’s fastest supercomputer, to the United States Department of Energy (DOE) and the Lawrence Livermore National Laboratory (LLNL). This system, with a performance of 1.742 exaflopsAn exaflop is a unit of measure in supercomputing that …, not only ranks as the leader on the TOP500 list of the most powerful supercomputers but also holds a position among the top 20 most energy-efficient in the Green500 ranking.
A Leap in Performance and Energy Efficiency
El Capitan, which operates with a 100% fanless liquid cooling system, achieves an energy efficiency of 58.89 gigaflops per watt, underscoring its commitment to sustainability and technological innovation. Its computing capacity is equivalent to performing more than one quintillion operations per second, establishing it as a key resource for advancing AI-driven scientific research and strengthening the national security of the United States.
Trish Damkroger, Senior Vice President and General Manager of HPE’s HPC and AI Infrastructure Solutions, highlighted the significance of this achievement: “El Capitan marks a significant milestone in exascale computing, providing monumental performance and capabilities that will accelerate AI-driven scientific discoveries, enhance national security, and open new opportunities in renewable energy”.
A Strategic Ally for National Security
El Capitan is an essential tool for the National Nuclear Security Administration (NNSA), enabling advanced 3D simulations and the modernization of the U.S. nuclear arsenal without the need for physical testing. Additionally, the system will support secondary missions related to nuclear non-proliferation, counter-terrorism efforts, and research in areas such as materials discovery and high-energy-density physics.
Rob Neely, Director of the Weapons Simulation and Computing Program at LLNL, stated: “This resource will allow us to perform the high-fidelity simulations we need to fulfill our national security mission. Moreover, we will invest in developing AI on El Capitan to make our computations faster, more efficient, and potentially more accurate”.
Cutting-Edge Technology
The performance of El Capitan is based on an innovative platform, powered by the HPE Cray EX supercomputing solution and the AMD Instinct™ MI300A accelerator units, which integrate CPU and GPU cores with high-speed memory into a single package. The system also utilizes the HPE Slingshot interconnect, a high-speed Ethernet-based fabric, enabling massive computations across its more than 11,000 nodes.
Additionally, in collaboration with LLNL, HPE developed a customized local storage solution located near the nodes, which reduces latency and optimizes performance through a global Lustre file system shared by all compute nodes.
Commitment to Sustainability
El Capitan is the first supercomputer to implement a 100% fanless direct liquid cooling architecture, which not only increases its density and processing capabilities but also significantly reduces its energy consumption. This approach positions El Capitan at the forefront of sustainable computing systems, aligning with global priorities for energy efficiency and emission reduction.
Advancements in Scientific and Technological Research
Beyond its applications in national security, El Capitan will be utilized for research projects in areas such as climate change, modernization of electrical grids, nuclear fusion research, and drug discovery. This system, designed to combine traditional simulation workloads with AI-driven applications, promises to break new ground in scientific innovation.
With El Capitan, HPE reaffirms its leadership in the supercomputing sector, having developed the world’s only three exascale systems: El Capitan, Frontier, and Aurora, all of which occupy the top positions on the TOP500 ranking. This achievement reinforces HPE’s position as a driver of technological innovation in the era of artificial intelligence.