Skip to main content
Logo
The European High Performance Computing Joint Undertaking (EuroHPC JU)

Our supercomputers

Today the EuroHPC JU has procured nine supercomputers, located across Europe.

LUMI

386.00 petaflops
Sustained performance
539.13 petaflops
Peak performance

Compute partitions:

GPU partition (LUMI-G), x86 CPU-partition (LUMI-C), data analytics partition (LUMI-D), container cloud partition (LUMI-K)
Central Processing Unit (CPU): The LUMI-C partition features 64-core next-generation AMD EPYC™ CPUs
Graphics Processing Unit (GPU): LUMI-G based on the future generation AMD Instinct™ GPU
Storage capacity:

LUMI’s storage system consists of three components. First, there is a 7-petabyte partition of ultra-fast flash storage, combined with a more traditional 80-petabyte capacity storage, based on the Lustre parallel filesystem, as well as a data management service, based on Ceph and being 30 petabytes in volume. In total, LUMI has a storage of 117 petabytes and a maximum I/O bandwidth of 2 terabytes per second.

Applications:  AI,  especially deep learning, and traditional large scale simulations combined with massive scale data analytics in solving one research problem
TOP500 ranking: #1 in EU; #5 globally (November 2023)
Green500 ranking: #4 in EU; #7 globally (November 2023)
Other details:

LUMI takes over 150m2 of space, which is about the size of a tennis court. The weight of the system is nearly 150 000 kilograms (150 metric tons)

For information about pay per use conditions, please contact the hosting site directly: CSC IT - Centre for Science

Access LUMI now!

LEONARDO

246.54 petaflops
Sustained performance
313.90 petaflops
Peak performance
Compute partitions: 

GPU-partition (Booster) delivering 240 petaflops

x86 CPU-partition (Data-Centric) delivering 9 petaflops and featuring DDR5 Memory and local NVMe.

Central Processing Unit (CPU): Intel Ice-Lake (Booster), Intel Sapphire Rapids (Data-Centric)
Graphics Processing Unit (GPU):

13824 "Da Vinci" GPUs (based on NVIDIA Ampere architecture) delivering up to 10 exaflops of FP16 Tensor Flow AI performance

Storage capacity: Leonardo is equipped with over 100 petabytes with state-of-the-art hard disk drives and 5 petabytes with full flash and NVMe technologies.
Applications: The system targets: modular and scalable computing applications, data analysis, as well as interactive, urgent and cloud computing applications
TOP500 ranking: #2 in EU; #6 globally (November 2023)
Green500 ranking: #10 in EU; #18 globally (November 2023)
Other details: Leonardo is hosted in the premises of the Tecnopolo di Bologna. The area devoted to the EuroHPC Leonardo system includes 1240 sqm of computing room floor space and 900 sqm of ancillary space.

For information about pay per use conditions, please contact the hosting site directly: CINECA

Access LEONARDO now!

MARENOSTRUM 5

178.30 petaflops
Sustained performance
311.95 petaflops
Peak performance
Compute partitions:  GPP (General purpose partition), ACC (Accelerated partition), NGT GPP (Next Generation Technology General Purpose partition and NGT ACC (Next Generation Technology General Purpose partition). Additional smaller partitions for pre- and post-processing.
Central Processing Unit (CPU): The GPP, ACC partitions both rely on Intel Sapphire Rapids CPUs. NGT ACC is based on Intel Emerald Rapids and the NGT GPP is based on NVIDIA Grace.
Graphics Processing Unit (GPU): The ACC partition is based on NVIDIA Hopper whereas the NGT ACC partition is built on Intel Rialto Bridge.
Storage capacity: MareNostrum storage provides 248PB net capacity based on SSD/Flash and hard disks, and an aggregated performance of 1.2TB/s on writes and 1.6TB/s on reads. Long-term archive storage solution based on tapes will provide 402PB additional capacity. Spectrum Scale and Archive will be used as parallel filesystem and tiering solution respectively.
Applications:

All the applications suit ideally MareNostrum5 thanks to its heterogeneous configuration, with a special focus on medical applications, drug discovery as well as digital twins (earth and human body), energy, etc. Its large general-purpose partition provides an environment well suited for most current applications that solve scientific/industrial problems. In addition, the accelerated partition provides an excellent environment for large scale simulations, AI and deep learning.

TOP500 ranking: #3 in EU; #8 globally (November 2023)
Green500 ranking: #3 in EU; #6 globally (November 2023)
Other details:

MareNostrum 5 is located in BSC’s new facilities, next to the Chapel which is hosting previous systems. The datacenter has a total power capacity of 20MW, and cooling capacity of 17MW, with a PUE below 1,08.

For information about pay per use conditions, please contact the hosting site directly: Barcelona Supercomputing Center (BSC)

Access MARENOSTRUM 5 now!

MELUXINA

12.81 petaflops
Sustained performance
18.29 petaflops
Peak performance
Compute partitions: Accelerator - GPU (500 AI PetaFlops), Cluster (3 PetaFlops peak), Accelerator - FPGA and Large Memory Modules
Central Processing Unit (CPU): AMD EPYC
Graphics Processing Unit (GPU): NVIDIA Ampere A100
Storage capacity:

20 PetaBytes main storage with all-flash scratch tier over 600GB/s, Tape archival capabilities

Applications: AI, Digital Twins, Traditional Computational workloads, Quantum simulation
TOP500 ranking: #19 in EU; #71 globally (November 2023)
Green500 ranking: #14 in EU; #27 globally (November 2023)
Other details: Modular Supercomputer Architecture, Cloud Module for complex use cases and persistent services, Infiniband HDR interconnect, high speed links to the RESTENA NREN and GÉANT network, Luxembourg Internet Exchange and Public Internet

For information about pay per use conditions, please contact the hosting site directly: LuxProvide

Access MELUXINA now!

KAROLINA

9.59 petaflops
Sustained performance
12.91 petaflops
Peak performance
Compute partitions:

The supercomputer consists of 6 main parts:  

  • a universal part for standard numerical simulations, which will consist of approximately 720 computer servers with a theoretical peak performance of 3.8 PFlop/s,
  • an accelerated part with 70 servers and each of them being equipped with 8 GPU accelerators providing a performance of 11 PFlop/s for standard HPC simulations and up to 150 PFlop/s for artificial intelligence computations, 
  • a part designated for large dataset processing that will provide a shared memory of as high as 24 TB, and a performance of 74 TFlop/s,
  • 36 servers with a performance of 131 TFlop/s will be dedicated for providing cloud services,
  • a high-speed network to connect all parts as well as individual servers at a speed of up to 200 Gb/s,
  • data storages that will provide space for more than 1 PB of user data and will also include high-speed data storage with a speed of 1 TB/s for simulations as well as computations in the fields of advanced data analysis and artificial intelligence.
Central Processing Unit (CPU): More than 100,000 CPU cores and 250 TB of RAM
Graphics Processing Unit (GPU): More than 3.8 million CUDA cores / 240,000 tensor cores of NVIDIA A100 Tensor Core GPU accelerators with a total of 22.4 TB of superfast HBM2 memory
Storage capacity: More than 1 petabyte of user data with high-speed data storage with a speed of 1 TB/s
Applications: Traditional Computational , AI, Big Data
TOP500 ranking: #32 in EU; #113 globally (November 2023)
Green500 ranking: #13 in EU; #25 globally (November 2023)

For information about pay per use conditions, please contact the hosting site directly: IT4Innovations National Supercomputing Center

Access KAROLINA now!

DISCOVERER

4.52 petaflops
Sustained performance
5.94 petaflops
Peak performance
Compute partitions: One partition providing 1128 nodes, 4,44 petaflops
Central Processing Unit (CPU): AMD EPYC 7H12 64core, 2.6GHz, 280W (Code name Rome)
Graphics Processing Unit (GPU): No
Storage capacity: 2 petabytes
Applications: Traditional Computational, HPC as a Service / Federated HPC Supercomputing services
TOP500 ranking: #53 in EU; #166 globally (November 2023)
Green500 ranking: #81 in EU; #216 globally (November 2023)
Other details: Topology - Dragonfly+ with 200Gbps (IB HDR) bandwidth per link

For information about pay per use conditions, please contact the hosting site directly: Sofia Tech Park

Access DISCOVERER now!

VEGA

6.92 petaflops
Sustained performance
10.05 petaflops
Peak performance
Compute partitions:

CPU partition: 960 nodes with 2CPUs and 256GB memory/node (20% 1TB/node), 1x HDR100 & GPU partition: 60 nodes with 2CPUs and 512GB memory, 2x HDR100, 4x Nvidia A100/node

Central Processing Unit (CPU) : 2040x CPUs AMD EPYC 7H12 (64c, 2.6-3.3GHz), 130.560 cores on CPU and GPU partition
Graphics Processing Unit (GPU):

240x Nvidia A100 with 40 GB HBM2 (+4 on GPU login nodes), 6912 FP32 CUDA cores and 432 Tensor cores per GPU

Storage capacity: High-performance NVMe Lustre (1PB), large-capacity Ceph (23PB)
Applications: Traditional Computational, AI, Big Data/HPDA, Large-scale data processing
TOP500 ranking: #60 in EU; #198 globally (November 2023)
Green500 ranking: #90 in EU; #253 globally (November 2023)
Other details:

6x 100 Gbit/s bandwidth for data transfers to other national and international computing centres, data processing throughput of more than 400GB/s with high-performance storage and 200GB/s with large-capacity storage

For information about pay per use conditions, please contact the hosting site directly: IZUM

Access VEGA now! 

DEUCALION

7.22 petaflops
Sustained performance
10.00 petaflops
Peak performance
Compute partitions: ARM Partition: 1632 nodes, 3.8 PFLops ; x86 Partition: 500 nodes, 1,62 PFLops ; Accelerated: 33 nodes, 1,72 PFLops
Central Processing Unit (CPU):

A64FX (ARM partition), AMD EPYC (x86 partitions)

Graphics Processing Unit (GPU): NVidia Ampere
Storage capacity:

430 TB High-speed NVMe partition, 10.6 PB high-speed based Parallel File System partition.

Applications: Traditional Computational, AI, Big Data

For information about pay per use conditions, please contact the hosting site directly: MACC

JUPITER

1 Exaflop*
Sustained performance
Compute partitions:

Booster Module (highly-scalable GPU accelerated)

Cluster Module (general-purpose, high memory bandwidth)
Central Processing Unit (CPU):

The Cluster Module will utilise the SiPearl Rhea1 processor (ARM, HBM), integrated into the BullSequana XH3000 platform.

Graphics Processing Unit (GPU): The Booster Module will utilise NVIDIA technology, integrated into the BullSequana XH3000 platform.
Storage capacity:

JUPITER will provide a 20-petabyte partition of ultra-fast flash storage. The spinning disk and backup infrastructure capacity will be procured separately and subject to change.

Applications: JUPITER will be designed to tackle the most demanding simulations and compute-intensive AI applications in science and industry. Applications will include training large neural networks like language models in AI, simulations for developing functional materials, creating digital twins of the human heart or brain for medical purposes, validating quantum computers, and high-resolution simulations of climate that encompass the entire Earth system.

*Expected sustained performance

All other systems display the real performance of the combined partitions and are ordered according to the last Top 500 ranking.