The module system

Using software in modules¶

To run software on the cluster, it must be available in the system. Many common programs are pre-installed, but if a specific program is missing, you can install it in your home directory. For installation support, please submit a support request.

We use the Environment Modules system, allowing users to list available software and dynamically load it into their session. The command for this is module (or ml). See man module for details.

If you do not load a specific software version, the system will use the most recent available version.

Important: If the required software is not available, avoid installing it directly with pip. Instead, use a virtual environment or a container for proper installation.

We offer a module system with a lot of pre-installed software. If your software isn't available, we offer several compilers for compiling the software yourself. Many languages also have systems in place for user local installations, see Python, Perl below. If it's a common software, we can do a system installation.

You can also use libraries, build tools, and compilers from the module system to build your own software.

We use EasyBuild to build/install all software systematically.

Modules¶

A module system is used to allow the user to specify which application/software they want to use on the cluster. The modules package allows dynamical modification of a user's environment (PATH etc.), which makes the program available from the shell. Modules can be loaded and unloaded dynamically.

The `module` command¶

The module command (or the ml shortcut) allows you to manage available software:

man module # View the manual

To search for a module names:¶

module spider softwarename
ml spider softwarename

To search for keywords in a module (such as extensions in bundles):¶

module keyword foo
ml keyword foo

To view available software possible to load directly:¶

module avail
ml avail

To see what modules are loaded at the moment use:¶

module list
ml

To load/unload a module:¶

module load <modulename>
module unload <modulename>
module purge    # Unloads all currently loaded modules`
ml <modulename>
ml unload <modulename>
ml purge

Viewing Module Details:¶

To see what a module does, use

module show <modulename> # Display module environment variables
ml show <modulename>

Toolchains¶

A toolchain is a set of modules that makes up the necessary components to build most software.

Most commercial (closed source) tools, do not require a toolchain, e.g. Mathematica, MATLAB. These can be loaded directly;

module load MATLAB/2024a

other software is build for a particular toolchain, predominatly foss (GCC, OpenMPI, OpenBLAS), and intel.

graph TD
    GCCcore --> GCC & intel-compilers & NVHPC
    GCC --> gompi & gfbf --> foss
    intel-compilers --> iimpi & iimkl --> intel
    NVHPC --> nvompi --> nvofbf

Note: You can see the CUDA version in as a version suffix, indicating GPU support (but remember to look at the job statistics to verify that you are actually using the GPUs).

Finding compatible software¶

You can't mix toolchains; something built with e.g. foss-2024a can not be combined with anything built for foss-2023b. Loading most modules will load many additional modules which it depends on, and trying to swap them out will not be compatible. If you need a library updated for a newer toolchain you can request it by contacting the support.

The module command will complain if you accidentally try to load incompatible modules.

module load matplotlib/3.9.2-gfbf-2024a
module load CMake/3.27.6-GCCcore-13.2.0 # This will fail; it uses a different compiler version

To find which compiler you have currently loaded (via matplotlib/3.9.2-gfbf-2024a here) you can list them

module load matplotlib/3.9.2-gfbf-2024a
moduel list  # shows all the loaded dependencies

Where you will see it was GCCcore-13.3.0. You can also refer to the table below.

If incorrect modules are loaded, reset with:

module purge

Recent toolchains¶

Key software in recent toolchains:

Release	GCC	OpenMPI	Intel	CUDA	Python
2021a	10.3.0	4.1.1	2021.2.0	11.3.1	3.9.5
2021b	11.2.0	4.1.1	2021.4.0	11.4.1	3.9.6
2022a	11.3.0	4.1.4	2022.1.0	11.7.0	3.10.4
2022b	12.2.0	4.1.4	2022.2.1	12.0.0	3.10.8
2023a	12.3.0	4.1.5	2023.1.0	12.1.1	3.11.3
2023b	13.2.0	4.1.6	2023.2.1	12.4.0	3.11.5
2024a	13.3.0	5.0.3	2024.2.0	12.6.0	3.12.3