MareNostrum 5
Pre-Exascale supercomputer, co-funded by EuroHPC JU and the Mare Nostrum 5 consortium, led by Spain and which also includes Portugal and Turkey.
The system, hosted and operated by the Barcelona Supercomputing Center (BSC), is based on the Bull Sequana XH3000 solution supplied by Eviden and on Lenovo ThinkSystem architectures, with a maximum performance of 314 Petaflops, or 314 million billion calculations per second.
It is ranked #8 on the Top500 list of most efficient supercomputers, and #15 in the ranking of the world's greenest. Its architecture, equipped with liquid cooling systems and energy-efficient HPC technologies, makes it an extremely green supercomputer, despite its impressive computing power.
European researchers and users from the public sector and industry will be able to access this new system via the national Calls (CPCAs) and via Calls managed by the EuroHPC JU.
MareNostrum 5
Operational from 2024
MareNosctrum 5 technical specifications
Pre-Exascale co-financed by EuroHPC
Bull SAS combination of Bull Sequana XH3000 and Lenovo ThinkSystem architectures
GPP (General purpose partition), ACC (Accelerated partition), NGT GPP (Next Generation Technology General Purpose partition) and NGT ACC (Next Generation Technology General Purpose partition)
GPU: The ACC partition is based on NVIDIA Hopper, while the NGT ACC partition is built with Intel Rialto Bridge.
Storage with a capacity of 248PB based on SSD/Flash and hard disks, with aggregate performance of 1.2TB/s in writes and 1.6TB/s in reads.
295.81 petaflops maximum processing capacity
MareNostrum infographics 5
See all the information about the MareNostrum 5 cluster
Partitions
MareNostrum 5 provides users with the following partitions:
- GPP (General purpose partition), ACC (Accelerated partition)
- NGT GPP (Next Generation Technology General Purpose partition
- NGT ACC (Next Generation Technology General Purpose partition)
- Additional smaller partitions for pre- and post-processing.
MareNostrum 5 GPP (General Purpose Partition)
GPP-HighMem Nodes Equipped with:
ConnectX-7 NDR200 InfiniBand (shared by 2 nodes, 100Gb/s bandwidth per node)
2x Intel Xeon Platinum 8480+ 56C 2GHz
16x DIMM 16GB/64GB 4800MHz DDR5
960GB NVMe local storage
Additionally, GPP-HMB Nodes with:
2x Intel Xeon CPU Max 9480 56C 1.9GHz
8x Die 16GB 3200MHz HBM2
960GB NVMe local storage
ConnectX-7 NDR200 InfiniBand (shared by 2 nodes, 100Gb/s bandwidth per node)
MareNostrum 5 ACC (Accelerated Partition)
Equipped with:
2x Intel Xeon Platinum 8460Y+ 40C 2.3GHz(80 cores per node)
4x NVIDIA Hopper H100 64GB HBM2
16x DIMM 32GB 4800MHz DDR5(512GB main memory per node)
480GB NVMe local storage
4x ConnectX-7 NDR200 InfiniBand(800Gb/s bandwidth per node)
Software
Storage & Network
Access
Calls of Advanced Computing Projects
The RNCA organizes Calls for Advanced Computing Projects with the aim of allocating computing resources for research and innovation.
EuroHPC
Access to the EuroHPC portion of Deucalion is governed by the EuroHPC Joint Undertaking Access Policy. Information on access can be found on the dedicated EuroHPC Access page Calls .