The new system, known as Panther, consists of 32
nodes and 16 terabytes of RAM. The system will have 64 nVIDIA K80s which
makes this a very capable GPU facility.
Initially the Panther system will be open to just STFC and IBM
projects whilst we develop novel software to make effective use of the
performance potential of the system. However, the plan is to make the
new cluster available to industry clients and other research groups in
Work is currently underway on the installation of management
tools, storage and application software and integration into the user
authorisation system. Panther will become available to its first group
of users over the next few weeks, and an expansion to the system to form
a more powerful cluster is planned for later in 2016.
For more information contact Robin Pinning, Chief Technology Officer