LLNL debuts trio of systems on new Top500 list of world’s most powerful supercomputers, including El Capitan Early Delivery System

Newswise — Three new systems currently or soon-to-be sited at Lawrence Livermore National Laboratory (LLNL) on Monday debuted on the latest Top500 list of most powerful supercomputers in the world, including the first portion of the exascale machine El Capitan. Unveiled at the International Supercomputing Conference in Hamburg, Germany, the June 2024 Top500 lists three systems with identical components — one computing rack each from El Capitan’s “Early Delivery System” (EDS), LLNL’s newest unclassified supercomputer RZAdams and its unclassified “sister” system Tuolumne. All three registered 19.65 petaFLOPs (nearly 20 quadrillion floating point calculations per second) on the High Performance Linpack (HPL) benchmark used by the Top500 organization to determine the world’s fastest supercomputers. The scores ranked them 46^th, 47^th and 48^th in the world, respectively.

Each machine comprises the same HPE Cray EX supercomputer from Hewlett Packard Enterprise (HPE), featuring AMD architecture and nodes. The compute nodes are powered by the cutting-edge AMD Instinct^TM MI300A accelerated processing units (APUs), which integrate a tightly coupled central processing unit (CPU) and graphics processing unit (GPU) into a single package.

The APUs are expected to provide exceptional floating-point performance and artificial intelligence (AI)-assisted data analysis for the power required, enabling scientists from the National Nuclear Security Administration’s (NNSA) Tri-Labs (LLNL, Los Alamos and Sandia National Laboratories) to run high-resolution 3D models quicker and repeatedly, making those complex multi-physics simulations both routine and truer to life. All three systems also incorporate the open, Ethernet-based HPE Slingshot interconnect, which enables the entire supercomputer to act as one system with high-speed networking.

LLNL brought the El Capitan EDS cabinet on-line as part of the overall installation process of NNSA’s first exascale supercomputer El Capitan, which is projected to exceed 2 double precision exaFLOPs (2 quintillion operations per second) of peak performance, making it likely the world’s most powerful supercomputer when fully deployed. The EDS result constitutes an early test of El Capitan’s performance. Installation of El Capitan’s compute nodes began in March and remains ongoing, keeping the machine on schedule for initial use by NNSA Tri-Lab application teams later this year.

“We’re excited to be making significant progress on El Capitan and moving a step closer to harnessing the extraordinary power of NNSA’s first exascale supercomputer here at LLNL,” said LLNL Weapon Simulation and Computing Associate Director Rob Neely. “This is a tangible sign of advancement towards the promise of groundbreaking achievements in scientific research and national security, and we remain on track for deployment as a critical resource for the NNSA Tri-Labs beginning this fall.”

El Capitan will be used by the Tri-Labs for applications supporting NNSA’s stockpile modernization programs, as well as its stewardship mission to ensure the safety, security and reliability of the nation’s enduring nuclear stockpile in the absence of underground nuclear testing. It will also spur advancements in inertial confinement fusion energy, high energy density physics, material discovery, nuclear data, material equations of state and conventional weapon design.

Over the coming months, LLNL will complete installation of the remaining compute nodes and equipment, run a rigorous series of hardware, software, and system tests and formally accept the system.

As El Capitan’s largest unclassified companion system, Tuolumne shares the same architecture and APUs as El Capitan but is a fraction of the exascale machine’s size. Installation on Tuolumne (named after a river and a meadow in Yosemite National Park), began in late February at HPE’s Chippewa Falls facility. It is expected to be delivered to LLNL and deployed later this year, at roughly the same time as El Capitan, where it will support open science projects in material discovery, energy security, climate change, drug discovery, inertial confinement fusion, astrophysics, and other areas of public interest.

The third new system on the list, RZAdams, is a recently deployed unclassified system sited at LLNL. With compute nodes and infrastructure identical to El Capitan’s, the 19.65 PF peak performance benchmark makes it LLNL’s most capable unclassified machine, placing just ahead of Sierra’s companion Lassen.

Funded by LLNL, RZAdams was released to NNSA Tri-Lab users earlier this year, where it has been leveraged by application teams for critical software development and testing of a subset of the applications and libraries that will eventually run on El Capitan. One of the applications is MARBL, where it has been successfully used to model inertial confinement fusion experiments conducted at LLNL’s National Ignition Facility. RZAdams represents the Tri-Lab’s most powerful platform currently available to users for code porting and initial optimization for El Capitan.

"RZAdams is a critical addition to LLNL's existing and highly capable unclassified computing ecosystem,” said LLNL’s Chief Technology Officer for Livermore Computing Bronis R. de Supinski. “Given the boost in performance we are already seeing from the AMD Instinct MI300A APUs, we expect this system to be a strategically significant resource and to move us forward across a wide range of scientific areas. RZAdams is supporting essential opportunities for code porting, optimization and software development for some of the applications that will eventually run on El Capitan. It is also allowing our scientists to push the boundaries of discovery with greater speed and efficiency than previous unclassified systems.”

The addition of the three systems brings the total number of Top500 supercomputers sited or soon to be sited at LLNL to 14, far and away the most of any known computing institution in the world.

The number one system in the world remains Oak Ridge National Laboratory’s exascale machine Frontier, which slightly raised its benchmark HPL score from the Nov. 2023 list to 1.206 exaFLOPs. Argonne National Laboratory’s Aurora also officially became the world’s second exascale supercomputer, placing #2 on the list and reaching 1.012 EFs on the HPL.

For more, visit https://top500.org/.

AMD, the AMD logo, AMD Instinct, and combinations thereof are trademarks of Advanced Micro Devices, Inc.