The RCC Spear Cluster

The FSU Research Computing Center (RCC) maintains a cluster of 22 compute nodes, known as "Spear", for interactive computing and visualization. Eight of the Spear nodes are connected to GPU accelerators.

Any user with an RCC account can log in directly to a Spear node to run jobs interactively, using a command like:

        ssh -Y 
(The "-Y" or "-X" option facilitates X-window connections.)

When a user accesses Spear, a round robin process is used to select the specific node which will be used, to avoid overload on any particular node.

Each node on the cluster has an Infiniband connection to the Lustre parallel file system, the same file system used by the RCC high performance computing (HPC) cluster.

The Gunzburger group owns 8 of the Spear nodes, each of which has 16 Intel Xeon E5-2670 processors, with 2.6 GHz clock speed and 20 Megabytes of cache memory. The group also owns 1 Terabyte of Lustre file storage. Researchers using the Gunzburger Spear nodes regularly run Matlab, parallel Matlab, programs written in C, Fortran, and Python, and parallel programs making use of MPI and OpenMP.

Users who are authorized to access the Gunzburger Spear nodes can log in to those nodes specifically with a command like:

        ssh -Y
(The "-Y" or "-X" option facilitates X-window connections.)

If you are authorized to use the Gunzburger Spear nodes, you have a choice of using the general access nodes or the Gunzburger nodes. In theory, you might prefer the Gunzburger nodes, assuming they are less heavily used because fewer people have access to them. Of course, this is not always the case, and you might just find in some cases that it makes sense to switch over to the general access nodes if you find the Gunzburger nodes heavily loaded.

Although you should not normally need to worry about this, the Gunzburger nodes have local addresses spear-23 through spear-30.

Currently, there is not a job scheduler available on the Spear nodes, but one will be installed at the next upgrade, in July 2016.

For more information, see the RCC Spear website: .

You can return to the HTML web page.

Last revised on 27 May 2016.