About the System
The cluster has a head node and 30 computing nodes. Each node contains two 6-core Intel Xeon X5660 CPUs, and 48 GiB memory. Two nodes also have two Nvidia Tesla M2070 GPGPU each. The theoretical computing performance of the system (Rmax) is about 4+2 TFLOP/s.
The head node has 34 TB effective disk space, which is available from each node through 10Gib/s Ethernet network. Temporary data can be stored on the additional 500GB disks in each compute node.
Parallel programs' communication is on the low latency, 40 Gib/s wide InfiniBand network. The execution is scheduled by HTCondor.