Cluster Access

Princeton Clusters

Logging In

When using Tiger3, you should always use our dedicated tiger-arrk login node (see  🚢⁠⁠Tiger-ARRK⁠ ) rather than the general-access tiger3 login node.
You can access the clusters via SSH from the terminal. If connecting to a cluster via SSH is new to you, read the Princeton Research Computing " Connect by SSH " guide. To summarize, here are several addresses for the clusters we use:
Cluster
SSH
Tiger
ssh <NetID>@tiger-arrk.princeton.edu
Della
ssh <NetID>@della.princeton.edu
Della (GPU)
ssh <NetID>@della-gpu.princeton.edu
Neuronic
ssh <NetID>@neuronic.cs.princeton.edu
Adroit
ssh <NetID>@adroit.princeton.edu
For several clusters, there are also web browser-based options via the "OnDemand" service:

GUI Applications

To view and interact with GUI applications on the cluster, you need to support X11 forwarding. If you are on Windows and using  MobaXterm , everything should "just work" as-is. Similarly, if you're on Mac, you can install  XQuartz  and simply run the program before SSHing. Additional options can be found in the link below:

Transferring Files

To access and/or transfer files on an HPC machine, you have several options. The first is to use the  OnDemand  web browser interface to the cluster described above. However, it is often convenient to have an option that works independent of the browser, which we describe below.

SFTP Client

SFTP clients are programs that allow you to view and transfer files between your local machine and the clusters. For Windows, I strongly recommend  WinSCP . Mac folks tend to use  Cybderduck . Additional options recommended by Princeton Research Computing can be found below. You can also transfer individual files using the scp command-line function if desired.

VS Code

Another option is to edit and transfer files using VS Code via its  Remote - SSH  extension. If you have completed the  😴⁠⁠Removing Tedium⁠  guide, then you should be able to use VS code to remotely edit files on the clusters without issue! The VS Code option generally makes the most sense when you are writing code or working on a program.