Instructions for Lab Machine Access:

  1. Create a CS Account: If you don't already have one, register for a CS account here and list Julia as your sponsor. Approval from the CRF may take a few days.

  2. Notify Julia: Once your CS account is active (or if you already have one), please inform Julia of your CS account ID so she can add you to the Speech Lab server group, enabling access to the lab machines. Your CS account ID is typically the same as your UNI.

  3. Contact Lin for Machine Assignment: After Julia approves your access, reach out to Lin (lin.ai@cs.columbia.edu) to be assigned a specific machine. If your project requires extensive use of GPUs, request a GPU-equipped machine. Keep in mind that our GPU resources are limited, so opt for a regular CPU machine if possible. If you do need GPU, you may be assigned to one of our GPU machines or HPC clusters.

  4. [Optional] HPC Cluster: If you are assigned to an HPC cluster such as Ginsburg, Julia and Lin will facilitate access through the HPC support group. This process is typically quick, often completed in less than a few days.

Connecting to Lab Machines:

  1. Accessing a Regular CPU Machine: Use SSH to connect to your assigned CPU machine:

    ssh [your CS account ID]@[machine name].cs.columbia.edu
    
  2. Accessing a GPU Machine (Hecate or Kedi):

  3. Local Directory Usage: Store and run your projects in your local directory: /home/[account ID].

  4. Shared Resources:

  5. File Sharing and Management:

Connecting to HPC

  1. The only active HPC right now is Ginsburg. Use SSH to access Ginsburg:
ssh [your UNI]@ginsburg.rcs.columbia.edu
  1. Create a directory named with your UNI under /burg/katt3/users/. Only use /burg/katt3/users/[your UNI] to store your data and projects.
  2. For running jobs on Ginsburg, checkout the manual here.