Getting Started with Great Lakes


Important Notes


Creating an ARC account

  1. Request an HPC login here

    Please put “Accessing Great Lakes for the course “Music and AI” (PAT 498/598, Winter 2025)” in the request.

    Account Request

  2. Log into your account in the web portal

    You must connect to U-M VPN to access Great Lakes if you’re not connected to MWireless.

    Dashboard


Launching a Session

  1. Navigate to “Interactive Apps” in the web portal
  2. Select the application

    You’ll likely want to select “JupyterLab” for the assignments.

    Navigation -- Interactive Apps

  3. Specify the configuration of the instance (see below for suggestions)

    If you’re unsure, select “python3.11-anaconda/2024.02” as your Python distribution.

    Also, make sure to select the correct slurm account so that the cost is charged to the right account.

    Configuration

  4. Connect to the launched instance

    Interactive Sessions


Configuration

Configuration

Alternative configuration for a GPU Machine

Another alternative configuration for a GPU Machine

If you need more CPU cores on the standard partition, launch an instance with N cores with Nx7 GB of memory to maximize what you get.

Other configurations

For each GPU requested, Great Lakes provides a certain number of CPU cores and RAM memory without additional charge. That’s why we need to configure the instance properly so that we are not getting less CPU and memory than charged.

Here is the configuration of each partition (as of Winter 2025; see the current configurations here):

Partition CPU cores Memory (RAM) GPU GPU speed GPU memory #GPUs available
spgpu 4 48 GB A40 faster 48 GB 224
gpu_mig40 8 124 GB A100 fastest 40 GB 16
gpu 20 90 GB V100 fast 16 GB 52
standard* 1 7 GB - - - -

Here is the cost for each partition (as of Winter 2025; see the current rates here):

Partition Hourly cost CPU hours equivalent
spgpu 0.11 7.33
gpu_mig40 0.16 10.66
gpu 0.16 10.66
standard* 0.015 1

Storage


Tips

Copied from https://sled-group.github.io/compute-guide/great-lakes

Sometimes you want to quickly launch a node and ssh into it instead of launching a whole JupyterLab session or remote desktop. In that case, you can put tail -f /dev/null as the last command of your job, which will prevent the job from exiting without eating up CPU cycles. For example, your job script might be something like:

#!/bin/bash
echo $SLURMD_NODENAME
tail -f /dev/null

Then, either use the web interface or inspect the $SLURMD_NODENAME environment variable to figure out the node name and simply ssh into it from your login machine.

If you prefer to interact with Great Lakes using command line, you might find this cheat sheet helpful.


Hosted on GitHub Pages. Powered by Jekyll. Theme adapted from minimal by orderedlist.