March 27, 2024, on the Venado supercomputer at TA-3 at Los Alamos National Laboratory. Photo provided by LANL
LANL news release
In a collaboration to help advance national security and basic science missions, Los Alamos National Laboratory is working with Hewlett Packard Enterprise (HPE) and NVIDIA to design and install the institute's newest supercomputer, Venado. Ta. Representatives from the U.S. Department of Energy, National Nuclear Security Administration, Los Alamos, HPE, and NVIDIA showed off the tape of his new HPE Cray EX supercomputer Venado, equipped with the NVIDIA GH200 Grace Hopper superchip, at the Nicholas C. Metropolis Center on Monday. I made the cut. For modeling and simulation in the laboratory.
“Venado adds to the state-of-the-art supercomputing that advances national security and basic research, accelerating the way we integrate artificial intelligence to address these challenges,” said Los Alamos Director Tom Mason. says.
The NVIDIA Grace Hopper Superchip combines an Arm-based NVIDIA central processing unit with an NVIDIA Hopper architecture-based graphics processing unit for high-performance computing and large-scale artificial intelligence applications. Superchips can typically execute millions of instructions per second or more at lower cost and lower power consumption than traditional chip technologies. In initial scale testing, Venado has shown significant results in materials science atomic simulations and high-resolution astrophysical simulations.
“Our supercomputing capabilities represent a critical element in how national laboratories address important problems,” said U.S. Deputy Secretary of Energy David Turk. “With its ability to incorporate artificial intelligence approaches, we look forward to seeing how Los Alamos’ Venado system can deliver new and meaningful results in areas of interest.”
Venado's computing power includes an exascale-class HPE Cray EX supercomputer with 2,560 direct water-cooled Grace Hopper superchips. This system also uses 920 of his NVIDIA Grace CPU superchips and will be the first large-scale system equipped with his NVIDIA Grace CPU superchips to be deployed in the United States. NVIDIA packs 144 Arm cores on its Grace CPU superchip to instantly boost performance for a wide range of HPC applications.
“The Venado system, America's first supercomputer powered by NVIDIA Grace Hopper, delivers breakthrough performance and energy efficiency to accelerate scientific discovery,” said Ian Buck, vice president of hyperscale and HPC at NVIDIA. says Mr. “Through his continued work with Alamos and his HPE, Venado will become a great scientific instrument for researchers to achieve breakthroughs in materials science, renewable energy, astrophysics, and more. ”
The Venado system was built to prioritize overall performance and workflow efficiency. The system is networked with HPE Slingshot 11 interconnects, and the extremely fast network significantly improves overall system performance. Additional his HPE Cray supercomputing software will improve modeling and simulation workloads and enable researchers to leverage his AI and machine learning to leverage insights and make new discoveries. Venado is designed and manufactured at HPE's facility in Chippewa Falls, Wisconsin, and takes its name from one of the highest peaks in New Mexico's Sangre de Cristo Mountains. This system was delivered to Los Alamos in his March.
Trish Damkroger, Senior Vice President and General Manager, HPC & AI Infrastructure Solutions, HPE . “Los Alamos National Laboratory continues to demonstrate cutting-edge research and engineering results with industry-leading supercomputing. We are proud of our long-standing relationship with the Institute, which provides innovative AI solutions that accelerate discovery and make impactful contributions in the field.”
Venado is the result of a co-design process that forms the basis of an ongoing collaboration focused on developing a wide range of computing, memory, and software technologies. Codesign combines the expertise of vendors, hardware architects, system software developers, domain scientists, computer scientists, and applied mathematicians to collaborate and make informed decisions about hardware and software components.
Venado supercomputer installed at Los Alamos. Photo provided by LANL
