The Physics and Astronomy Department (DIFA) hosts a parallel computing infrastructure directly accessible to its staff members. This infrastructure is made up of three different groups (or islands) of computing nodes with different hardware features and target usage:

• BladeRunner: a set of compute nodes targeted to low-parallelism jobs (with hostnames bldNN)

• Matrix: a set of compute nodes targeted to massively parallel jobs (with hostnames mtxNN)

• GPU: a set of compute nodes with Graphic Processing Units besides standard CPUs (with hostnames gpuNN)

For a more detailed description of the hardware features and access policies of the different groups of nodes see the Resources section.

The access to the infrastructure> is provided by frontend servers, that can be used to connect and transfer data to the cluster using the standard protocols ssh, scp, rsync.