Systems Information

Skip to end of metadata
Go to start of metadata

Clusters

CaSToRC manages several HPC clusters with a total Theoretical Peak Performance of ~40TFlops and aggregated storage of ~400TB

CyTera

CyTera is a hybrid CPU/GPGPU Linux cluster. There are 98 CPU only compute nodes and a further 18 computes nodes that also have 2 GPUs each.

  1. Compute Node specifications
    Node Type Node Name Model CPU RAM Disk Other
    compute n001-n098 IBM dx360 M3 2 x XeonX5650 (Westmere hexacore) @ 2.67GHz 48GB DDR3 @1333MHz 146GB NA
    gpucompute g01-g18 IBM dx360 M3 2 x XeonX5650 (Westmere hexacore) @ 2.67GHz 48GB DDR3 @1333MHz 146GB 2 x Nvidia M2070 (Tesla) GPUs
  2. Interconnect (MPI Message Passing and Storage Access Network)
    Mellanox 4xQDR Infiniband (40Gbps)
  3. Storage
    300TB GPFS

Prometheus

Prometheus is a CPU/GPGPU Linux cluster. There are 8 compute nodes that each have 2 GPUs.

  1. Compute Node specifications
    Node Type Node Name Model CPU RAM Disk Other
    gpucompute node01-node08 IBM dx360 M3 2 x XeonX5650 (Westmere hexacore) @ 2.67GHz 24GB DDR3 @1333MHz 146GB 2 x Nvidia M2070 (Tesla) GPUs
  2. Interconnect (MPI Message Passing and Storage Access Network)
    Mellanox 4xQDR Infiniband (40Gbps)
  3. Storage
    1TB disk on Management node

Euclid

  1. Compute Node specifications
  2. Interconnect (MPI Message Passing and Storage Access Network)
  3. Storage

Planck

  1. Compute Node specifications
  2. Interconnect (MPI Message Passing and Storage Access Network)
  3. Storage

Storage

Labels:
None
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.