Los Alamos taps NVIDIA Vera CPUs for new supercomputers

Los Alamos National Laboratory is preparing three new supercomputers, Mission, Vision and Veritas, with HPE and NVIDIA to accelerate scientific discovery and support agentic AI for science. The systems will use HPE Cray Supercomputing GX5000 architecture with the NVIDIA Vera Rubin platform, combining NVIDIA Vera CPUs, NVIDIA Rubin GPUs and NVIDIA Quantum-X800 InfiniBand networking.

Mission’s planned configuration includes NVIDIA Vera Rubin GPU nodes and 2,300 standalone NVIDIA Vera CPUs on the HPE Cray Supercomputing GX240 blade. Veritas will add approximately 1,150 standalone NVIDIA Vera CPUs alongside NVIDIA Vera Rubin nodes and will support the Laboratory Directed Research and Development program. Vision will serve fundamental science, including materials and nuclear science, energy modeling, biomedical research and AI.

LANL is using agentic workflows in which AI agents form hypotheses, choose tools, launch simulations, analyze results and refine next steps. Its URSA framework is designed to help scientists brainstorm hypotheses, plan experiments, run simulations and analyze results, running on Venado and soon Mission and Vision. LANL demonstrated the Vera CPU delivered 7x higher performance on URSA workloads than CPUs in the Crossroads x86 supercomputer.

Early LANL testing on Branson, an open source Monte Carlo heat transfer simulation tool, showed Vera outperforming Crossroads x86 CPUs by over 3x. A single Vera CPU outperforms a single socket x86-based CPU by more than 3x while providing more than 4x the memory per core and 6x the memory per node. Mission and Vision are expected to be operational in 2027, extending LANL and NVIDIA’s collaboration from Grace to Vera.