PFE Usage Guidelines

Article ID: 181
Last updated: 06 Jan, 2026

The PFEs are the front-end systems for Aitken, Electra, and Endeavour. They provide an environment that enables quick turnaround for tasks such as file editing, file transferring, compiling, and short debugging/testing sessions, as well as for batch job submissions via PBS to Aitken, Electra, to Endeavour.

WARNING: The PFEs use Intel Sandy Bridge processors. If you use a PGI compiler to build your executable on the PFEs, be aware that by default the executable is optimized for the Sandy Bridge microarchitecture (which includes Sandy Bridge and Ivy Bridge) and will not necessarily execute on the compute node processors. See PGI Compilers and Tools for information on generating a single executable that will work on all processor types.

You cannot use SSH to access the compute nodes except for the subset of nodes your PBS job is running on.

Restrictions on PFEs

Default usage quotas are established on the PFEs to support the availability and reliability of the these systems for all users. If you exceed the default usage limits, the memory and processor resources available to you on the PFE will be reduced for a period of time, and you will be notified via email.

The hard quota defaults for each user are 4 cores for processor usage and 8 gigabytes (GB) for memory usage. The soft quota defaults for each user are 2 cores for CPU usage and 4 GB of memory usage.

For details, see Usage Quotas on PFEs.

WARNING: MPI (Message Passing Interface) jobs are not permitted to run on the PFEs.

Pre-Processing and Post-Processing Data

For pre- and post-processing applications such as Tecplot, IDL, and MATLAB, we recommend using a reserved front end (RFE) or the Lou data analysis nodes (LDANs).

You can reserve an RFE by running the "reservable front end" command (pbs_rfe), specifying what type of node you want and for how long. The command makes a PBS advanced reservation for a node and tells you which node is reserved.

You can request the LDANs in the qsub command line. The Lou home filesystems, NAS compute resources home filesystems (/homeX/username), and /nobackup filesystems are all mounted on the LDANs.

Note: There are only four LDAN nodes, so there are often wait times for using them.

File Transfers to Mass Storage

The /nobackup filesystems are mounted on Lou, so the easiest way to transfer files between NAS compute systems and Lou is to initiate a command such as shiftc, cp, mcp, or tar on Lou. For example:

lou% shiftc /nobackup/username/filename $HOME

If you initiate the transfer from a PFE, you can use the commands scp or shiftc to transfer files between a PFE and Lou. For very large file transfers, we recommend the NAS-developed Shift tool (shiftc).

File transfers from the compute nodes to Lou must first go through one of the PFEs.

When sending data to Lou, keep your largest individual file size under 1 terabyte (TB). Files larger than 1 TB will occupy all of the tape drives, preventing other file restores and backups.

Also read

Contact Us

General User Assistance

Security

  • Report security issues 24x7x365
  • Toll-free: 1-877-NASA-SEC (1-877-627-2732)
  • E-mail: soc@nasa.gov

User Documentation

High-End Computing Capability (HECC) Portfolio Office

NASA High-End Computing Program

Tell Us About It

Our goal is furnish all the information you need to efficiently and effectively use the HECC resources needed for your NASA computational projects.

We welcome your input on features and topics that you would like to see included on this website.

Please send us email with your wish list and other feedback.