hpclab.petnica.rs

Welcome to the official landing page of the Petnica Science Center High Performance Computing Labaratory. On this page you can find information about the work and equipment of the Labaratory.

HPC Lab is managed by volunteers from the Department of Computer Science in Petnica Science Center.

Cluster Specifications

Compute Nodes

  • c1: Intel® Core™ i7-5820K (6c/12t), 64GB RAM, 1× NVIDIA GTX 960 4GB (second card withdrawn, hardware fault)
  • c2: Intel® Core™ i7-5820K (6c/12t), 64GB RAM, 2× NVIDIA GTX 960 4GB
  • c3: Intel® Core™ i7-5930K (6c/12t), 64GB RAM, 2× NVIDIA GTX 960 4GB
  • c4: Intel® Core™ i9-7900X (10c/20t), 64GB RAM, 2× NVIDIA GTX 1060 6GB
  • c5: Intel® Core™ i7-5820K (6c/12t), 64GB RAM, 1× NVIDIA GTX 970 4GB (second card withdrawn, hardware fault)
  • Total: 5 nodes, 34 cores / 68 threads, 320GB RAM, 8 GPUs

Head Node

  • head1: Intel® Core™ i7-5820K (6c/12t), 16GB RAM — login, scheduling and file server (no compute GPU)

Storage

  • Shared Home: 1.9TB SSD/NVMe RAID10, exported to every node over NFS
  • Local Disk: 110GB – 445GB SSD per compute node (system)

Network

  • Interconnect: 1GbE Ethernet

Software

  • Scheduler: Slurm Workload Manager
  • Operating System: Rocky Linux 10.2
  • GPU Driver: NVIDIA 580.173.02 (CUDA compute capability 5.2 and 6.1)

Acknowledgements

Some of the hardware the cluster runs on is donated. We thank the companies below for making the HPC Lab possible.

Nordeus

Nordeus donated the machines used for the head node (head1) and the compute nodes c1–c5. The default SLURM partition is named nordeus in their honour.