The European supercomputer Leonardo, managed by Cineca, that will be installed at the end of 2022 in the new data center located in the Tecnopolo of Bologna, will be capable of an aggregated HPL performance of 250 Pflops (250 trillion operations per second) and equipped with over 100 petabytes of state-of-the-art storage capacity. Moreover, it will deliver 10 exaflops of FP16 AI performance.  the system will provide 10 times the computing power of Cineca’s current top tier system Marconi100, which is currently ranked in the ninth position on the global TOP500 list of the world’s most powerful supercomputers. 

    EuroHPC JU 

    Leonardo is the first of three pre-exascale systems announced by EuroHPC, a collaboration between national governments and the European Union. Funded by the European Commission and by the Italian Ministry of Universities and Research, EuroHPC’s aim is to develop a world-class supercomputing ecosystem and exascale supercomputing in Europe. The other pre-exascale class systems, that will join the Italian Leonardo will be installed in Finland and Spain. 

      Cineca's objective is to create a European reference for the technological supervision and development of supercomputing, which is a strategic asset for the independence and competitiveness of Italy and Europe as a whole. 

        HPC in Italy and in Europe 

        Italy has defined a national high performance computing roadmap, and has given a strong mandate to Cineca to act as hosting entity in agreement with the Italian Ministry of Universities and Research, the National Institute of Nuclear Physics (INFN) and the International School of Advanced Studies (SISSA), and in collaboration with the Emilia-Romagna Region for the location of the data center. The Hosting Entity Consortium sees also the participation of Slovenia, Slovakia, Austria, Hungary, and soon also Greece. 

        Italy has a long, successful tradition in the field of supercomputing, thanks to the foresight of the Ministry of Universities and Research. Moreover, Cineca's strong operational capacity and expertise act as a guarantee to the national scientific community, in addition to a European commitment for a suitable and competitive supercomputing infrastructure. 

        High-performance computing, along with advanced algorithms, data and hardware based on the next generation of exascale supercomputers, is strategic to tackle forthcoming challenges in science, technology and industry, and to help public decision-making during natural risk events or societal challenges. In recent years, great strides have been made by the United States, China and Japan in this area, with supercomputing considered to promote the economic success of countries. Through the initiative of the EuroHPC Joint Undertaking, Europe has decided to invest in European exascale advanced computing infrastructures to meet the growing computing and data needs of European scientists and industry. 

        The research challenges of the future 

        The Leonardo system will contribute at the mitigation and management of risks due to extreme situations, natural events, earthquakes, tsunamis, volcanic events, hydrogeological assets, for the fight against pandemic situations and pathogen epidemics. In fact, Italy is leading some of the European reference projects on combating COVID-19, such as the Escalate4CoV project for the reuse of known and synthetic active ingredients in the preparation of new drugs for contrast therapy against COVID, as well as the Orchestra project, for the management and monitoring of contagion clusters to prevent the spread of the pandemic. 

        Similar initiatives are intended to stimulate other priority issues that have emerged so far, which include quality of the environment, cybersecurity and AI. 

        The procurement

        Leonardo will be based on Atos BullSequana XH2000 technology, and feature nearly 14,000 next generation NVIDIA Ampere architecture-based GPUs and NVIDIA Mellanox HDR InfiniBand.

        The procurement was announced on October 15 2020 by the Italian Minister of Universities and Research Gaetano Manfredi, together with Roberto Viola, Director General of DG Connect at the EU Commission and Cineca President Eugenio Di Sciascio and General Director David Vannozzi. Also in attendance were by Thomas Skordas, director of "Digital Excellence and Science Infrastructure"DG Connect; Anders Dam Jensen, Executive Director of EuroHPC Joint Undertaking, Sanzio Bassini, Director of Cineca's HPC department, Antonio Zoccoli, INFN President, Stefano Ruffo, SISSA President, Giuseppe Di Franco, CEO of Atos Italia, and Marc Hamilton, vice president of solutions architecture and engineering at NVIDIA. 

        Technical Information 

        Leonardo will be built from Atos’ BullSequana XH2000 supercomputer nodes, each with four NVIDIA Tensor Core GPUs and a single Intel CPU. It will also use NVIDIA Mellanox HDR 200Gb/s InfiniBand connectivity, with smart in-network computing acceleration engines that enable extremely low latency and high data throughput to provide the highest AI and HPC application performance and scalability. 

        Leonardo will feature nearly 14 000 NVIDIA Ampere architecture-based GPUs. It will deliver 10 exaflops of FP16 AI performance. NVIDIA Ampere architecture GPUs can accelerate over 1,800 applications such as Quantum Espresso for material science, SPECFEM3D for geoscience and MILC for quantum physics by up to 70x, making previous big challenge simulations almost real-time tasks. 

        3rd Gen Intel Xeon Scalable processors (Ice Lake) are optimized to perform computationally intensive workloads in high-performance computing systems like Leonardo. The follow-on processor to Intel’s Ice Lake server processors is Sapphire Rapids, which will enable exascale computing with advanced built-in AI acceleration capabilities. 

        • More than 136 BullSequana XH2000 Direct Liquid cooling racks 
        • 250 PFLOPs HPL Linpack Performance (Rmax) 
        • 10 ExaFLOPS of FP16 AI performance
        • 3456 servers equipped with Intel Xeon Ice Lake and NVIDIA Ampere architecture GPUs 
        • 1536 servers with Intel Xeon Sapphire processors 
        • 5PB of High Performance storage 
        • 100PB of Large Capacity Storage