We offer every one of our researchers access to high performance workstations and compute servers. On the desktop we have a wide variety of equipment ranging from the top-end latest Mac Pros to multiple CPU linux machines with plenty of RAM. It's our goal to provide the finest personal computing environment to all our staff and faculty.
Supporting our dynamic, static and video capture systems are a series of powerful SAN arrays serving three quarters of a petabyte of disk. We're expecting to break the petabyte barrier in the coming year
Perceiving Systems shares a compute cluster with other departments in the Intelligent Systems Institute. The IS cluster offers approximately 40 compute nodes with a total of more than 1500 CPU cores and 15TB of RAM.
Our GPU cluster supports training large deep networks. It includes 42 Nvidia K20 GPUs and 40 K80 GPUs. This will continue to grow. Additionally researchers have access to the Max Planck supercomputing center in Garching, which has a GPU cluster with 676 Nvidia K20s.
The components are located in three different buildings and are connected via 1GBit and 10Gbit Ethernet components. The storage systems offer more than 200TB hard disk capacity.
A large percentage of these nodes are cooled with an advanced water-cooling system. Nodes run the latest stable version of the popular Ubuntu Linux OS, and jobs are submitted and controlled via the HTCondor.