How-To: Using the Cluster¶
This section will try to use some common situations to illustrate the basic means of accomplishing your work on the cluster, and pointing you to the various other resources we have on how to do this.
The main theme is roughly:
- Development and short-feedback iterative work should probably be Interactive Jobs. This is the cluster’s equivalent to working on your laptop/desktop to figure things out.
- When you want to run your software for longer, or in greater numbers, or unattended then you want to look at Batch Jobs. This is the main usage of the cluster, and where people spend the bulk of their time.
- Batch Jobs
- Interactive Jobs
- Array Jobs
- Parallel Jobs
- GPU Computing
Doing Large I/O, or Downloading large files¶
Again do this via, Interactive Jobs instead of on the login nodes.
Please read the section storage which outlines the types of storage systems on the cluster and which one you should be using for your task.