Los Alamos National Laboratory will implement NVIDIA Vera CPUs in new supercomputers, enhancing agentic AI capabilities for scientific research. This development aims to accelerate the formation of hypotheses and data analysis in scientific projects.
Los Alamos National Laboratory (LANL) plans to develop new supercomputers named Mission, Vision, and Veritas, in collaboration with HPE and NVIDIA. These systems will utilize the HPE Cray Supercomputing GX5000 architecture, incorporating NVIDIA Vera CPUs alongside NVIDIA Rubin GPUs. The focus is on enhancing the capabilities of agentic AI for scientific research.
Early testing of the NVIDIA Vera CPUs demonstrated a significant performance increase, achieving 7x better results on URSA workloads compared to the CPUs in the Crossroads x86 supercomputer. Additionally, Vera CPUs surpassed Crossroads CPUs by over 3x in the Branson simulation tests. These improvements are attributed to features such as the custom Olympus core and advanced memory handling in Vera CPUs.
The new supercomputers will facilitate the development of AI agents capable of forming hypotheses, selecting tools, and running simulations. This aligns with LANL's goals to enhance scientific discovery through automated processes, spearheaded by the development of their Universal Research and Scientific Agent (URSA), which will operate on these advanced systems. The goal is to streamline the scientific process from hypothesis formulation to data analysis.
The hardware and software for the new supercomputers have been co-designed by experts across multiple disciplines including domain scientists and computer scientists. This collaborative approach ensures that the systems are tailored to meet the demands of actual scientific workloads, rather than just optimizing for synthetic benchmarks.
The Mission supercomputer is projected to be operational by 2027, representing the fifth Advanced Technology System developed at LANL. This system will play a crucial role in advancing the lab's research objectives and enhancing the role of AI in scientific inquiry.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
Los Alamos National Laboratory will implement NVIDIA Vera CPUs in new supercomputers, enhancing agentic AI capabilities for scientific research. This development aims to accelerate the formation of hypotheses and data analysis in scientific projects.