Software

As much fun as echoing "Hello World" is, it's hard to perform analyses without software. We've seen that software packages are not available on the login nodes, but now that we're interactively connected to a compute node, we can view and use what's available.

Since the HPC is a shared system, we need to organize our installed packages in a way that's helpful for everyone. There are three main categories of software on HPC:

System Software

System software represents executables pre-installed on the system, and are available everywhere. This includes basic bash commands, as well as tools developed by our team, like va and nodes-busy.

Modules

Modules are software packages that are only accessible on compute nodes. Modules make it easy to load and unload software from your environment and allow hundreds of packages to be available on the same system without dependency or versioning conflicts.

Specifying module versions

It's always good practice to specify which software version you need when loading a module to ensure a stable environment.

You can view and load software modules using the commands module avail and module load, respectively. For example:

[netid@gpu66 ~]$ module avail python

------------------------------------------------- /opt/ohpc/pub/modulefiles --------------------------------------------------
   python/3.6/3.6.5    python/3.8/3.8.2 (D)    python/3.8/3.8.12    python/3.9/3.9.10

[netid@gpu66 ~]$ module load python/3.9
[netid@gpu66 ~]$ python3 --version
Python 3.9.10

Try running module avail without specifying any arguments. You'll notice we have a lot available. These software packages can be used both interactively for shorter jobs and testing, or in your batch scripts.

If you have software that you would like installed as a module, please refer to our software policies to see if it's a good candidate.

Personal Software

Installing software packages locally into your own directories is encouraged. This means you can download, install, manage, and customize software from the internet (e.g. Pip, CRAN, GitHub) without waiting for someone else to install it on HPC for you. Software should be compiled on a compute node so that the load doesn't strain the login nodes and to ensure you have access to the newest system software and compilers.

If you would like to install your own software, please refer to our guidelines in User Installations which also has some handy tips and tricks if you've never installed software locally (i.e., not in a root-owned location) before.