TLDR:
- Los Alamos National Laboratory partners with HPE and Nvidia to launch supercomputer ‘Venado’.
- Venado is equipped with Nvidia Grace Hopper Superchips and HPE Cray EX supercomputer.
The collaboration between Los Alamos National Laboratory, Hewlett Packard Enterprise (HPE), and Nvidia has resulted in the launch of the supercomputer ‘Venado’. Venado, housed at the Nicholas C. Metropolis Center for Modeling and Simulation, is powered by Nvidia Grace Hopper Superchips and the exascale-class HPE Cray EX supercomputer. The system includes 2,560 direct, liquid-cooled Grace Hopper Superchips and 920 Nvidia Grace CPU Superchips, making it the first large-scale system with Nvidia Grace CPU superchips in the United States. Venado is designed to prioritize overall performance and workflow efficiency, and it is networked with HPE Slingshot 11 high-speed interconnect. The supercomputer has shown promising results in atomistic simulations for materials science and high-resolution astrophysics simulations.
Venado’s combination of Arm-based Nvidia central processing units and Nvidia Hopper architecture-based graphics processing units allows it to address high-performance computing and massive-scale artificial intelligence applications, performing millions of instructions per second at a lower cost and power consumption compared to previous technologies. It is expected that Venado’s integration of artificial intelligence approaches will lead to new and impactful results across various areas of research.
Key figures involved in the project, including Thom Mason, director of Los Alamos National Laboratory, David Turk, deputy secretary of the U.S. Department of Energy, Ian Buck, vice president of hyperscale and HPC at Nvidia, and Trish Damkroger, senior vice president and general manager of HPC & AI Infrastructure Solutions at HPE, have expressed their optimism about Venado’s capabilities to accelerate scientific discovery and contribute to breakthroughs in fields like materials science, renewable energy, and astrophysics.
The development of Venado is the result of a codesign process involving collaboration from vendors, hardware architects, system software developers, domain scientists, computer scientists, and applied mathematicians. This collaboration has laid the foundation for ongoing work focused on advancing computing, memory, and software technologies for future supercomputing projects.