LEONARDO

The pre-exascale Tier-0 EuroHPC supercomputer LEONARDO is classified in 6° position among the most powerful supercomputers in the  Top500 List. It is hosted by Cineca and is currently built in the Bologna Technopole. It is supplied by ATOS, with two main partitions: Booster Module and Data-centric Module. The booster module partition is based on BullSequana XH2135 supercomputer nodes, each with four NVIDIA Tensor Core GPUs and a single Intel CPU. It also uses 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. The Data-centric partition is based on BullSequana X2140 three-node CPU Blade and is equipped with two Intel Sapphire Rapids CPUs, each with 56 cores.

LEONARDO will be available in pre-production from February 2023 and will be open to full production in March 2023. For more news and information please visit Leonardo's website.

System Architecture

Compute Nodes:

• 5092 computing nodes subdivided in:
  • • Model:  BullSequana X2135 "Da Vinci" single node GPU Blade
      Nodes: 3456 booster nodes
      Processors: Intel Xeon 8358 32 cores, 2.6 GHz 
      Cores: 110592 (32 cores/node)
      Accelerators: 4XNvidia custom Ampere GPU 64GB HBM2
      RAM: (8x64) GB DDR4 3200 MHz
      Network: 2xNvidia HDR cards 2x100Gb/s
    • Model:  BullSequana X2140 three-node CPU Blade
      Nodes: 1536 data-centric nodes
      Processors: Intel Saphire Rapids 2x56 cores, 4.8 GHz 
      Cores: 172032 (112 cores/node)
      RAM: (48x32) GB DDR5 4800 MHz 
      Network: 3xNvidia HDR cards 1x100Gb/s
• 16 visualization nodes  2 x Icelake ICP06 32cores 2.4GHz, 3 NVIDIA Tesla V100, RAM: (16 x 32) GB DDR5 4800 MHz
• 106 PB (raw) Large capacity storage, 620 GB/s
• High Performance Storage 5.4 PB, 1.4 TB/s Based on 31 x DDN Exascaler ES400NVX2 

Login and Service nodes: 16 Login nodes are available. 16  service nodes for I/O and cluster management.

All the nodes are interconnected through an Nvidia Mellanox network, with Dragon Fl+, capable of a maximum bandwidth of 200Gbit/s between each pair of nodes.

User guide