El Capitan

Security defines Lawrence Livermore National Laboratory’s mission. It’s a sweeping responsibility, encompassing efforts from ensuring the safety of the U.S.’s nuclear deterrent to addressing threats posed to energy and environmental security.

For over 65 years, the lab has met this mission, applying science and technology to make the world safer. And since its inception, high-performance computing leadership has been fundamental to LLNL’s success.

Lawrence Livermore National Lab and Cray forged their initial HPC partnership in 1978 when the lab was among the first to install a Cray-1 supercomputer. Now, LLNL and Cray, a Hewlett Packard Enterprise company, start a new ascent with El Capitan — the lab’s first exascale system and the world’s fastest supercomputer. Based on the Shasta architecture developed by Cray and enhanced with a variety of new HPE technologies, El Capitan will run at a record-breaking speed of 2 exaflops.

Scheduled for delivery in early 2023, the system will become the U.S. Department of Energy’s third exascale-class supercomputer, joining its Shasta-based sisters Frontier and Aurora at the Oak Ridge and Argonne National Laboratories. It will feature a new system architecture, new interconnect, new software platform, and next-generation AMD EPYC™ processors and high-bandwidth memory (HBM)-enabled AMD Radeon™ Instinct GPUs.

If security defines LLNL’s mission, then El Capitan redefines how they meet it.

Read the announcement for details on the El Capitan system.

Cray’s new supercomputer, code-named Shasta, is an entirely new design and will underpin the next era of supercomputing. The system is characterized by exascale performance capability, new datacentric workloads, and an explosion of processor and accelerator architectures. Shasta incorporates next-generation Cray system software to enable modularity and extensibility, supports unparalleled flexibility in processing choice, and scales seamlessly.

The Cray-designed Slingshot™ interconnect is the network backbone for the diverse simulation, modeling, AI, and analytics workloads today’s scientific and technological questions require. It offers a host of important new features such as adaptive routing, congestion control, and Ethernet compatibility.

Cray’s new, open, and extensible software platform addresses the growing need for supercomputing capability across government and private industry. It fuses supercomputing performance and capability with the modularity, composability, and ease-of-use of cloud computing, effectively answering today’s new class of digital transformation problems. This next-generation supercomputing software platform tackles key requirements, including application development and portability, management and monitoring, interoperability, and investment protection.