The AMD-powered El Capitan supercomputer, housed at the Lawrence Livermore National Laboratory (LLNL), is now ranked as the world’s fastest supercomputer.
Built by HPE, the supercomputer uses AMD Instinct MI300A accelerated processing units (APUs). It achieved a High-Performance Linpack (HPL) score of 1.742 exaflops based on the latest Top500 list.
The LLNL is using the supercomputer for nuclear security. El Capitan is the first exascale-class machine for the National Nuclear Security Administration (NNSA) and will be used to advance scientific discovery and national security, providing what AMD says is “the computational power necessary to ensure the safety, security and reliability of the nation’s nuclear deterrent without testing”.
It is being used for modelling and simulation capabilities to support NNSA’s Stockpile Stewardship Programme, which certifies the ageing nuclear stockpile and other critical nuclear security missions, such as non-proliferation and counter terrorism.
“El Capitan is crucial to the National Nuclear Security Administration’s core mission and significantly bolsters our ability to perform large ensembles of high-fidelity 3D simulations that address the intricate scientific challenges facing the mission,” said Rob Neely, director of LLNL’s advanced simulation and computing programme.
LLNL and the other NNSA at Los Alamos and Sandia National Laboratories are also using El Capitan and its companion system, Tuolumne, to drive artificial intelligence (AI) and machine learning-assisted data analysis. El Capitan will apply AI to high energy density problems such as inertial confinement fusion research, while Tuolumne will be used for unclassified open science applications including climate modelling, biosecurity/drug discovery and earthquake modelling.
Bronis R de Supinski, LLNL’s chief technology officer for Livermore Computing, said: “With AI becoming increasingly prevalent in our field, El Capitan allows us to integrate AI with our traditional simulation and modelling workloads, opening new avenues for discovery across various scientific disciplines.”
AMD said its Instinct MI300X and MI325X accelerators provide AI performance and memory capabilities, while the AMD Instinct MI300A APU puts central processing unit (CPU) and graphics processing unit (GPU) cores and stacked memory together into a single package, enabling “new levels of efficiency and performance” for high-performance computing (HPC) and AI workloads.
Its EPYC processors and Instinct accelerators are also being used to power many new supercomputing and AI projects and deployments, including Italian energy company Eni, whose HPC 6 supercomputer is powered by AMD EPYC processors and AMD Instinct GPUs. The University of Paderborn is also set to take delivery of a new supercomputer powered by the latest fifth-generation AMD Epyc technology.
Separately, IBM and AMD have announced a collaboration to deploy AMD Instinct MI300X accelerators as a service on IBM Cloud. The new service, available in the first half of 2025, will target performance and power efficiency for generative AI models. Through the collaboration, support for AMD Instinct MI300X accelerators is being made available within IBM’s Watsonx AI and data platform, as well as through the Red Hat Enterprise Linux AI inferencing platform.