The FRCE cluster is a 3000+ core Linux cluster designed for large numbers of simultaneous jobs ranging from microscopy or sequence analysis to chemical modeling. A wide variety of scientific software is available  along with access to MySQL and Oracle databases. Persistent data storage through various NAS platforms allows group access to shared data. The system is open to all NCI researchers.

The login node (batch.ncifcrf.gov, also referred to as the head node) is used to submit jobs to the cluster. Users connect to this via ssh. No compute intensive or large data transfer processes should be run on this node. A hard limit of 10 CPU minutes is enforced on the login node and any process exceeding this limit will be terminated.

Four different classes of compute nodes are currently installed. CPU nodes are the general compute systems. GPU nodes have either Nvidia P100 or V100 gpu cards. The large-memory systems have 3TB ram each for memory-intensive jobs. Dragen servers are systems designed specifically for next-get sequencing and their use is restricted to specific groups. A more detailed description of the hardware capabilities is maintained here. The partitions to use to specify particular node types is documented on this page.

The file transfer nodes are the preferred systems for running large data transfers through scp or rsync. No compute-intensive jobs are allowed on these systems.

NCIF maintains a Globus server to facilitate automated or large data transfers. Data may be moved within NIH as well as with other Globus-connected sites.

Some web or application systems are allowed privileges to submit jobs to the FRCE cluster. These are set up on a case-by-case basis. Contact the HPC support staff if you have need to set up a web server with FRCE access.

NAS shares are connected to the FRCE systems through NFS protocols. Persistent storage can be requested from EIT services. It is not possible to mount Windows-only shares on or from  the FRCE cluster.