IT Infrastructures

Infraestructures

CiTIUS computing infrastructures are a tool for the research activity, the development of projects in collaboration with the industry and to support the centre's daily activity.

These infrastructures are housed in a modern Datacenter of 22 m² and 8 cabinets of 48U located in the basement of CiTIUS building and allow the centre to have the capacity to develop projects and services in the following lines:

  • Applications parallelisation
  • Optimising applications on multi-core architectures
  • Optimising applications through GPGPU computing
  • Design and development of cloud architectures

HPC Cluster

We have a heterogeneous computing cluster with 16 nodes on which the centre's researchers can run computationally expensive applications.

  • 9 Dell R740 nodes each with two Intel Xeon Gold 5220R 24-core processors and 192 GB of RAM.
  • 1 Dell R840 fat node with four Xeon Gold 6248 20-core processors and 1 TB of RAM.
  • 2 Dell R7525 nodes each with two AMD EPYC 7543 32-core processors, 256 GB of RAM and two NVIDIA Ampere A100 GPUs.
  • 2 Dell R740 nodes each with two Intel Xeon Gold 5220 18-core processors, 192 GB of RAM and two NVIDIA Tesla V100s GPUs.
  • 2 Dell R7725 nodes with two AMD EPYC 9255 processors, 384 GB of RAM and two NVIDIA L4 24 GB GPUs each.

The cluster has a high-performance distributed storage BeeGFS system with more than 200 TB of capacity interconnected by a dual 25Gb optical fibre network.

General Purpose GPU Computing

We have several servers equipped with different GPUS available for users who need that kind of infrastructure to run their programs or to develop applications under the GPGPU computing paradigm.

SIE LADÓN 4214 server (2 Intel Xeon Silver 4214, 192 GB RAM)

  • NVIDIA Quadro P6000 24GB
  • NVIDIA Quadro RTX8000 48GB
  • 2 NVIDIA A100 80GB

3 SIE LADÓN servers (2 AMD EPYC 9474F, 768 GB RAM each)

  • 4 NVIDIA H200 NVL 141GB

SIE HGX server (2 AMD EPYC 9335, 1536 GB RAM)

  • 4 NVIDIA H200 141GB

2 SIE LADÓN inference servers (2 AMD EPYC 9255, 384 GB RAM each)

  • 2 NVIDIA L40S 48GB

Virtual Private Servers Cloud (VPS)

We offer researchers an IaaS-type service based on Proxmox. It offers virtual private servers in the cloud running on top of five servers connected to a distributed storage. This provides users with a fully configurable environment for development, testing, deployments, etc.