clusters

This is an old revision of the document!


Clusters

The majority of HPC resources are organized in clusters of machines. These machines are based on the same type of technology that you can find in your desktop or laptop PC, but have more powerful CPUs, more RAM, and a faster network than you would normally find in a consumer level machine. In addition to the hardware, the clusters run software that manages the computational jobs - making sure that the load is distributed over the machines and so ensuring that jobs run quicker. Finally, scientific and engineering software is pre-installed on the systems so you can just run your problems. Avalanche

Avalanche is the newest cluster at FEUP, being purchased at the start of 2013. It has 29 nodes, each with 16 cores and between 64GB and 128GB of RAM. Each node is connected via a 40Gbps InfiniBand network. There is 10TB of central storage, with each node having 500GB of local storage. The clusters run version 7 of the Scientific Linux operating system (a distribution based on Red Hat Enterprise Linux 7) from CERN and Fermilab.

For more information on how to use this cluster, please see the documentation.

The Avalanche cluster was funded and by a consortium of interested parties who have heavy computational requirements, the members of the consortium are as follows:

The IDMEC cluster consists of 39 nodes. Each has 2 dual core AMD Opteron Processors and 4GB of RAM. The nodes are connected via a dedicated 1Gbps Ethernet network. The cluster has a central file server with a capacity of 2TB. It provides a shared user home directory across all of the cluster. Each cluster node also has 90GB of local storage for IO intensive tasks. The cluster runs the SuSE Enterprise 10 operating system. The system can be accessed via the SSH protocol via a central logon machine idmecluster.fe.up.pt. From this machine you can submit your jobs to run on the cluster and monitor their status.

The INEB cluster consists of 18 PowerPC based nodes with 4 CPUs each and 4GB of RAM. Every node is connected via a dedicated 1Gbps Ethernet network. The cluster has two central file servers with a combined storage capacity of 1TB, these provide shared user home directories across all of the cluster. Each cluster node also has 33GB of local storage for IO intensive tasks. The cluster runs the SuSE Enterprise 9 operating system. The system can be accessed via the SSH protocol via a central logon machine - inebcluster.fe.up.pt. From this machine you can submit your jobs to run on the cluster and monitor their status.

The CAS cluster is a group of 8 nodes each with 8 cores and 24GB of RAM. The machines are part of the IBM Portugal Center for Advanced Studies (CAS) in Enterprise Engineering and Management at FEUP.

  • clusters.1709305331.txt.gz
  • Last modified: 2024/03/01 16:02
  • by ptsilva