SCU Resources and Services

This list contains the resources and services provided by the SCU. To access them, you need an SCU account.

Getting Help with HPC

Search this Website:

This site contains handful of information. You can find helpful content by entering keywords into the search box located at the top right of the page.

Contact Scientific Computing Team

Please contact us at scu@med.cornell.edu for further assistance.

When reaching out to us, please include detailed information to help us quickly identify and resolve the issue or answer your question.

Support Request Guide

When emailing for support, please review the information below to ensure we can assist you efficiently. This should be used a guide, as it does not need to be in this exact format and not all questions may be applicable.

1. Cluster Name:

   • Indicate which cluster you're using.

2. Issue Description:

   • Provide a brief summary of the issue you're experiencing.

   • Did this work before, or is this your first time attempting it?

   • List the steps you've already tried to resolve the issue.

3. Details Section(if applicable):

   • Include any relevant IDs (e.g., job IDs) and specify the node(s) where the error occurred.

   • Attach screenshots of the steps or errors you encountered; small details can often be helpful.

   • Share the specific commands or steps you followed.

   • Include any error messages.

   • Provide the absolute path to any relevant output files or logs (e.g., SLURM - /path/to/output/file).

   • Mention whether you’ve tried searching for the error message or troubleshooting online.

4. Additional Information:

   • Include details about your environment (e.g. software versions).

FAQ

1. How do I access Scientific Computing HPC cluster?

   1. Request a SCU Account

   2. Use the form above to request access to the computational resources provided by the SCU. We require your institutional or departmental affiliation as well as the name of your administrator, or PI, to fulfill the request. Please include your PI in the email, as we would need their approval. 

2. I have an account but I’m getting a connection closed(off campus)

   1. Users are added to a ticket to obtain remote access over the VPN(completed by a different dept). If you’re still receiving the error, please email us at scu@med.cornell.edu.

3. Which partitions do I have access to?

   1. All users will have access to the general partitions on BRB/Cayuga cluster: scu-gpu and scu-cpu

4. Can you install XYZ software?

   1. Before reaching out to team members, please check on the cluster if the software is already installed.

   2. If a software is not available, please reach out to scu team members and we can install the software you: Software Requests

5. How do I check/load software modules?

   1. To check which software modules are available on the cluster, use:

   2. module avail

   3. To load a specific module:

   4. module load module_name

6. What should I do if my job is killed or runs out of memory?

   1. Jobs can be killed if they exceed the allocated resources. Check your job’s memory and CPU usage using:

   2. sacct -j job_id --format=JobID,MaxRSS,MaxCPU

7. Who can I contact for support?

   1. Always email scu@med.cornell.edu even if you email specific team members directly.

8. What should I do if I encounter a "Disk quota exceeded" error?

   1. This error indicates that you have exceeded your allocated disk space. You can check your disk usage and remove unnecessary files.

Software Requests

Requesting New Software to be Added Cluster-Wide

Step 1: Check Existing Software

Before requesting new software, ensure it is not already available by using the module avail commands.

Step 2: Submit a Request

If the software is not available, submit a request to the scientific computing team scu@med.cornell.edu. Include the following information in your request:

• Cluster location

• Software name and version with link

• Any specific configuration or dependencies required

Step 3: Follow Up

After submitting your request, follow up with the support team if necessary to check the status of your request.