User guide#
This section covers the core of using Sherlock day-to-day once you have an account and can connect to the cluster.
-
All computational work on Sherlock must go through the Slurm scheduler. Login nodes are shared and have strict resource limits; they are only suitable for lightweight tasks. This page covers how to request resources, write batch scripts, and submit and monitor your work.
-
Sherlock OnDemand provides a web-based interface to the cluster, accessible from any browser without a SSH client. Manage files, submit and monitor jobs, and launch interactive applications such as Jupyter notebooks, RStudio, or a full remote desktop session.
-
Sherlock includes nodes with GPUs and other specialized hardware for workloads that require them. This page covers how to request and use GPU nodes in your jobs.
-
When something isn't working as expected, this page covers how to diagnose common problems and how to submit an effective support request to the SRCC team.