resources-available.markdown

layout
default

Resources available

Compute nodes

The compute nodes are the physical machines where you can run Virtual Machines (VMs).

• 32 HPC compute nodes, each node having:

  • 64 CPU cores (hyperthreading enabled)
  • 256 GB RAM
  • 3.2 TB local SSD disk
  • 2x 10 Gbit network connection per compute node

• 12 GPU compute nodes, each node having:

  • 32 CPU cores
  • 2 GPU's of type K2 GRID, accessible through PCI passthrough, enabling applications to get the performance boost of the direct access to a GPU card
  • 256 GB RAM
  • 800 GB local SSD disk
  • 2x 10 Gbit network connection per GPU node

• 1 High memory node, having:

  • 40 CPU cores
  • 2 TB RAM
  • 3.2 TB local SSD disk
  • 2x 10 Gbit network connection

Storage resources

• 900 TB Ceph distributed object storage with SSD caching
  • 2.7 PB total shared storage, used in x3 redundancy, which results in 900 TB net storage available
  • 10 Gbit network connection per storage node
  • Ceph storage is accessible as a block device within a VM