EuroHPC Joint Undertaking has awarded a contract to Eviden for the LISA upgrade of the Leonardo supercomputer, enhancing its AI capabilities by integrating an optimized partition designed specifically for large language models and multi-modal generative AI. This upgrade includes 166 advanced GPU servers with high-performance networking and memory, significantly boosting performance for AI-intensive tasks. Leonardo, a pre-exascale system ranking ninth globally with 241.2 petaflops, will connect to IT4LIA in Bologna as part of a new AI hub. The deployment is scheduled for late 2025, marking the first EuroHPC partition tailored for AI workloads.
EuroHPC Joint Undertaking Awards Contract for LISA
The EuroHPC Joint Undertaking (EuroHPC JU) has awarded Eviden a contract for the LISA supercomputer upgrade, which will enhance Leonardo with an AI-optimized partition. This upgrade supports the development of large language models and multi-modal generative AI, which are crucial for advancing AI capabilities.
The LISA upgrade incorporates advanced components, including 166 GPU servers, high-performance networking, and high-bandwidth memory, significantly improving performance for AI-intensive tasks. As the first EuroHPC partition designed specifically for AI workloads, it sets a new standard in computational efficiency.
Leonardo and LISA are integrated through a shared storage layer, facilitating essential data services for AI tasks. This integration complements the deployment of IT4LIA, establishing Bologna as a state-of-the-art AI hub within Europe’s expanding AI ecosystem.
Leonardo, a pre-exascale supercomputer ranking 9th globally with a performance of 241.2 petaflops, is a cornerstone of EuroHPC JU’s initiative to support European research and industry. Its capabilities underscore the organization’s commitment to fostering innovation through high-performance computing.
Eviden was selected following a tender call in September 2024, ensuring the upgrade meets rigorous standards. The project will be managed by EuroHPC JU and CINECA, with deployment scheduled for late 2025, marking a significant step forward in Europe’s AI infrastructure development.
Details of the LISA Supercomputer Upgrade
The LISA supercomputer upgrade introduces an AI-optimized partition to Leonardo, enhancing its capabilities for developing Large Language Models and multi-modal generative AI. This upgrade includes 166 advanced GPU servers interconnected via a high-performance non-blocking network, alongside high-bandwidth memory, to improve performance for AI-intensive tasks.
Leonardo and LISA are integrated through a shared multi-protocol storage layer that supports concurrent block, file, and object functionalities. This integration ensures seamless data services essential for AI computing tasks, enabling efficient workflows and resource utilization. The deployment of IT4LIA, the EuroHPC AI Factory in Bologna, further complements this upgrade by establishing a state-of-the-art AI hub in Italy.
Deployment and Integration of LISA with IT4LIA
The deployment of LISA will integrate an AI-optimized partition into the Leonardo supercomputer, enhancing its capabilities for developing Large Language Models and multi-modal generative AI. This upgrade includes 166 advanced GPU servers interconnected via a high-performance non-blocking network, alongside high-bandwidth memory, to improve performance for AI-intensive tasks.
Leonardo and LISA are integrated through a shared multi-protocol storage layer that supports concurrent block, file, and object functionalities. This integration ensures seamless data services essential for AI computing tasks, enabling efficient workflows and resource utilization. The deployment of IT4LIA, the EuroHPC AI Factory in Bologna, further complements this upgrade by establishing a state-of-the-art AI hub in Italy.
The LISA supercomputer upgrade is scheduled for late 2025, with EuroHPC JU and CINECA overseeing its implementation. This project reflects Europe’s commitment to advancing high-performance computing and artificial intelligence infrastructure, supporting research and industry across the continent.
More information
External Link: Click Here For More
