LAMOS gets its own HPC resource

Andreas Hilboll, 01 Dec 2016, News


Last week, the new HPC cluster for the LAMOS group was installed in the University's GreenIT Housing Center. About 15 months after sending the initial proposal, we will now be able to put our very own HPC cluster into operation, relieving us from having to borrow compute time from the Burrows working group (thanks a lot for helping us out in the meantime!) at IUP.

The HPC cluster is called aether, and consists of 56 compute nodes with two Intel Xeon E5-2690v4 CPUs each, for a total of 1568 compute cores running at 2.6 GHz. The cluster has a total memory of ~9 TB (40 nodes of 128 GB and 16 nodes of 256 GB each), and is supported by 480 TB parallel BeeGFS storage. It has a total (theoretical) peak performance of 65.2 Tflop/s.

aether was delivered and installed by RAID Media Systems, and their engineers did a great job in setting everything up according to our needs - thanks a lot, guys!

We will give access to aether to all interested working groups, especially at MARUM and IUP; a separate announcement regarding the technical and administrative details will follow.