Accessing and Using the SJMC-GPU01 Remote System

This documents describes the method to connect to the remote GPU server SJMC-GPU01

SJMC-GPU01 is a remote system with the following specifications

Operating system: Ubuntu 22.04: Includes Lambda Stack for managing TensorFlow, PyTorch, CUDA, cuDNN, etc.
Processor: AMD Ryzen Threadripper PRO 5995WX: 64 cores, 2.7~4.5GHz, 256 MB cache, PCIe 4.0
GPUs: 2x NVIDIA RTX 6000 Ada Generation (EDU/Inception): 48GB memory, 18176 CUDA cores, 568 Tensor cores
System memory: 512 GB: DDR4-3200 ECC RDIMM
OS drive: 1x 3.84 TB M.2 NVMe
Data drive: 2x 15.36 TB U.2 NVMe: Data center SSD, 1 DWPD, PCIe 4.0

Requesting Access

Please email someone to request access...

Connection Information


  • Use an SSH client.
    • macOS users can use the built in Terminal application.
    • Windows users can use Putty or another SSH client.
    • Linux users can use their terminal application of choice.
  • Log in with your Netid:

Graphical Remote Desktop

  1. Use a Remote Desktop Connection client.
  2. Launch your Remote Desktop client and connect to
    • When creating your connection you can enter your UW NetID and password in the settings for the connection and you will not be prompted to log in to the remote system.

If you did not enter your credentials in the connection settings you will be greeted by a log in window.

login screen

Leave the Session drop down box at Xorg and enter your UW NetID and password.

You should now be at the desktop of the system.


Sessions persist after quitting your Remote Desktop client. You can start work, quit your connection, then come back later.

Connecting to Research Drive

Research Drive shared will auto mount from a terminal when you are signed in.

  1. Open a terminal
  2. cd to /mnt/researchdrive/DRIVE_NAME (usually a NetID)

Maintenance Times

Maintenance will be performed on Tuesday mornings, and may require a reboot which will end active sessions. Please plan your work accordingly and contact if you anticipate a job running during this time.

Keywordsremote sjmc gpu cccr mcrc   Doc ID130537
OwnerSterling A.GroupSchool of Journalism & Mass Communication
Created2023-08-23 09:58 CSTUpdated2023-08-24 07:00 CST
SitesSchool of Journalism & Mass Communication
Feedback  0   0