News Hub

El Capitan supercomputer will be 10 times as powerful as anything we’ve seen before

Written by Thu 5 Mar 2020

“The Captain” breaks supercomputing ground with 2 exaflop milestone and features AMD processors

The US Department of Energy’s (DoE) upcoming El Capitan supercomputer will be capable of 2 exaflops of computing performance, making it more powerful than the top 200 fastest supercomputers combined.

The record-breaking system is expected to be delivered in early 2023 and will be located at the Lawrence Livermore National Laboratory (LLNL) in California. It will be used by the DoE’s National Nuclear Security Administration to support the reliability and security of the US’s nuclear stockpile through AI-powered advanced simulation and modelling.

LLNL researchers will also use the system for high-performance AI and machine learning applications to aid medical and drug research initiatives, including cancer drug discovery and the analysis of proteins linked to human cancers.

HPE is optimising El Capitan for the tasks it will perform and the supercomputer is based on Cray’s Shasta architecture.

Although Shasta supports the use of different types of processors within the same system, the DoE has decided to use upcoming AMD CPUs and GPUs. The Radeon Instinct GPUs are based on a new architecture for HPC and AI workloads, while the Genoa CPUs will feature a “Zen 4” processor.

The department did not initially reveal which chipmaker would provide the processors when the supercomputer was announced last year, opting instead to wait until it could determine which chips would offer the best value for the performance required.

“The exceptional computing power promised by El Capitan, based on HPE’s Cray Shasta architecture, will ensure the NNSA laboratories can continue to excel at their national security missions and make it possible for the U.S. to remain competitive on the global stage in high-performance computing for many years to come,” said Bill Goldstein, director, Lawrence Livermore National Laboratory.

“We look forward to continuing our work with HPE and AMD to usher in this new exascale era with the most capable hardware on the planet,” he added.

Last month, the UK Met Office announced it will spend £1.2 billion over 10 years building a supercomputer to improve rainfall predictions and airport forecasting.

Written by Thu 5 Mar 2020


hpc supercomputer US
Send us a correction Send us a news tip