HPC Lab - High Performance Computing Lab
To support the numerical modelling research and consulting activities carried out by the HTC-Group and ARES group, a dedicated High Performance Computing (HPC) facility is available at the DIEF site in Calenzano. The HPC-Lab includes the following two Linux clusters:
HPC cluster «Roger»
- Installation Dec. 2019
- 1400 cores – Intel Xeon Gold 6248 2.5 GHz + AMD Epyc
- 6Tb total RAM (distributed, 28 nodes)
- 100 Gbit/s Infiniband interconnection
HPC cluster «Thor»
- Installation Dec. 2023
- 1500 cores –AMD Epyc + 2 Nvidia A100 GPU nodes
- 9Tb total RAM (distributed, 28 nodes)
- 200 Gbit/s Infiniband interconnection
Clusters are configured to execute multi-physics CFD based simulations using different codes ranging from commercial packages (Ansys® Fluent, CFX, Converge®, Comsol®) up to open source codes (OpenFOAM®) and in-house research codes.
The DIEF site in Santa Marta hosts the HPC infrastructure, a Linux cluster dedicated to research and technology transfer activities for the TGroup:
HPC Cluster "Scrat"
Installation: September 2024
32 AMD Epyc 32-Core 7543
1024 cores 2.8 GHz (256MB L3 Cache)
4 TiB RAM DDR4-3200
15 TB storage
2 AMD Epyc 32-Core 9334
64 cores 2.7 GHz (128MB L3 Cache)
0.5 TiB RAM DDR5-4800
2 NVIDIA H100-NVL 94GB
100 Gbit/s Infiniband interconnection
The cluster is configured for performing CFD simulations and in-house development of advanced codes in the fields of aerodynamics, aeromechanics, and noise prediction, with a specific focus on turbomachinery. Component optimization challenges are addressed using AI-based technologies, aiming for rapid practical applications and industrial relevance.
Last update
17.02.2026