Module system commands

Objectives

  • Find software module versions
  • Load a module (with potential prerequisites):
    • default (often not recommended!)
    • a specific version of software
  • Unload software modules, including specific versions
  • Unload all software modules with module purge
    • This should never be done at PDC!
    • You can do this at UPPMAX, but load the uppmax module afterwards.
  • List the loaded modules
  • Get some information about a module (module show and module help)
  • Learn about module collections to load/unload a bunch of modules (module save <collection> and module restore <collection>)

In the previous section we saw how to find out which modules are available at a centre, including looking for a specific software module.

Now it is time to learn more useful module commands!

Finding software versions

If you just load a module without specifying the version, you will (depending on centre) get the default version or an error.

There are good reasons not to just load the default version, even when it is possible:

  • The default version is often the newest one installed, which means the default changes the next time an even newer version is installed. This could break your code or at least give a different result, which is bad for resproducibility.
  • Sometimes you actually need a specific version, which could differ from the one the centre has decided on as default.

So how do you find out which versions are available for a specific software?

Note

This is again done differently, depending on the center you are at:

  • HPC2N, LUNARC, C3SE, PDC: module spider MODULE or ml spider MODULE
  • UPPMAX, NSC: module avail MODULE or ml avail MODULE

With module spider

Recommended at HPC2N, LUNARC, C3SE, and PDC.

Note

The command to find the available versions for a software module named MODULE is: 

module spider MODULE

or, in short form 

ml spider MODULE

With module avail

Recommended at UPPMAX and NSC.

Note

The command to find the available version for a software module named MODULE is: 

module avail MODULE

or, in short form 

ml av MODULE

Example

Finding available versions of Python.

Tip

Type along!

module spider

module spider Python
Click to show 
b-an01 [~]$ module spider Python

--------------------------------------------------------------------------------------------------------------------
  Python:
--------------------------------------------------------------------------------------------------------------------
    Description:
      Python is a programming language that lets you work more quickly and integrate your systems more effectively.

     Versions:
        Python/2.7.15
        Python/2.7.16
        Python/2.7.18-bare
        Python/2.7.18
        Python/3.7.2
        Python/3.7.4
        Python/3.8.2
        Python/3.8.6
        Python/3.9.5-bare
        Python/3.9.5
        Python/3.9.6-bare
        Python/3.9.6
        Python/3.10.4-bare
        Python/3.10.4
        Python/3.10.8-bare
        Python/3.10.8
        Python/3.11.3
        Python/3.11.5
        Python/3.12.3
     Other possible modules matches:
        Biopython  Boost.Python  Brotli-python  GitPython  IPython  Python-bundle-PyPI  bx-python  flatbuffers-python  ...

--------------------------------------------------------------------------------------------------------------------
  To find other possible module matches execute:

      $ module -r spider '.*Python.*'

--------------------------------------------------------------------------------------------------------------------
  For detailed information about a specific "Python" package (including how to load the modules) use the module's full name.
  Note that names that have a trailing (E) are extensions provided by other modules.
  For example:

     $ module spider Python/3.12.3
--------------------------------------------------------------------------------------------------------------------

module spider Python
Click to show 
[bbrydsoe@cosmos3 ~]$ module spider Python

----------------------------------------------------------------------------------------------------------------------------
  Python:
----------------------------------------------------------------------------------------------------------------------------
    Description:
      Python is a programming language that lets you work more quickly and integrate your systems more effectively.

     Versions:
        Python/2.7.18-bare
        Python/2.7.18
        Python/3.8.6
        Python/3.9.5-bare
        Python/3.9.5
        Python/3.9.6-bare
        Python/3.9.6
        Python/3.10.4-bare
        Python/3.10.4
        Python/3.10.8-bare
        Python/3.10.8
        Python/3.11.3
        Python/3.11.5
        Python/3.12.3
     Other possible modules matches:
        Biopython  GitPython  IPython  Python-bundle  Python-bundle-PyPI  bx-python  flatbuffers-python  graphviz-python  ...

----------------------------------------------------------------------------------------------------------------------------
  To find other possible module matches execute:

      $ module -r spider '.*Python.*'

----------------------------------------------------------------------------------------------------------------------------
  For detailed information about a specific "Python" package (including how to load the modules) use the module's full name.
  Note that names that have a trailing (E) are extensions provided by other modules.
  For example:

     $ module spider Python/3.12.3
 ----------------------------------------------------------------------------------------------------------------------------

module spider Python
Click to show 
[brydso@alvis1 ~]$ module spider Python

-----------------------------------------------------------------------------------------------------------------------------
  Python:
-----------------------------------------------------------------------------------------------------------------------------
    Description:
      Python is a programming language that lets you work more quickly and integrate your systems more effectively.

     Versions:
        Python/2.7.18-GCCcore-11.2.0-bare
        Python/2.7.18-GCCcore-11.3.0-bare
        Python/2.7.18-GCCcore-12.2.0-bare
        Python/3.9.5-GCCcore-10.3.0-bare
        Python/3.9.5-GCCcore-10.3.0
        Python/3.9.6-GCCcore-11.2.0-bare
        Python/3.9.6-GCCcore-11.2.0
        Python/3.10.4-GCCcore-11.3.0-bare
        Python/3.10.4-GCCcore-11.3.0
        Python/3.10.8-GCCcore-12.2.0-bare
        Python/3.10.8-GCCcore-12.2.0
        Python/3.11.3-GCCcore-12.3.0
        Python/3.11.5-GCCcore-13.2.0
        Python/3.12.3-GCCcore-13.3.0
     Other possible modules matches:
        Biopython  Boost.Python  CUDA-Python  GitPython  IPython  Python-bundle-PyPI  flatbuffers-python  meson-python  ...

-----------------------------------------------------------------------------------------------------------------------------
  To find other possible module matches execute:

      $ module -r spider '.*Python.*'

-----------------------------------------------------------------------------------------------------------------------------
  For detailed information about a specific "Python" package (including how to load the modules) use the module's full name.
  Note that names that have a trailing (E) are extensions provided by other modules.
  For example:

     $ module spider Python/3.12.3-GCCcore-13.3.0
-----------------------------------------------------------------------------------------------------------------------------

module spider, mostly

module spider python
Click to show - python 
bbrydsoe@login1:~> ml spider python
-----------------------------------------------------------------------------------------------------------------------------
  python:
-----------------------------------------------------------------------------------------------------------------------------
     Versions:
        python/2.7.6
        python/2.7.9
        python/2.7.11
        python/2.7.15
        python/3.3
        python/3.3.1
        python/3.4.3
        python/3.5.0
        python/3.6.0
        python/3.6.8
        python/3.7.2
        python/3.8.7
        python/3.9.5
        python/3.9.6
        python/3.10.8
        python/3.11.4
        python/3.11.6-gcc-oc3
        python/3.11.6-gcc-qyw
        python/3.11.6-gcc-spv
        python/3.11.8
        python/3.11.9-gcc-epm
        python/3.12.1
        python/3.12.3
        python/3.13.0-gcc-cgo
     Other possible modules matches:
        biopython  cray-python  dbus-python  pdc-python  py-meson-python  py-python-dateutil  python-2.7.18-gcc-11.2.0-namldiz  python-3.8.12-gcc-11.2.0-nwj732o  python-3.8.12-gcc-11.2.0-vhhulma  ...

-----------------------------------------------------------------------------------------------------------------------------
  To find other possible module matches execute:

      $ module -r spider '.*python.*'

-----------------------------------------------------------------------------------------------------------------------------
  For detailed information about a specific "python" package (including how to load the modules) use the module's full name.
  Note that names that have a trailing (E) are extensions provided by other modules.
  For example:

     $ module spider python/3.13.0-gcc-cgo
-----------------------------------------------------------------------------------------------------------------------------
Click to show - cray-python 
bbrydsoe@login1:~> module avail cray-python

------------------------------------ /opt/cray/pe/lmod/modulefiles/core -------------------------------------
   cray-python/3.10.10    cray-python/3.11.5 (D)    cray-python/3.11.7

  Where:
   D:  Default Module

If the avail list is too long consider trying:

"module --default avail" or "ml -d av" to just list the default modules.
"module overview" or "ml ov" to display the number of modules for each name.

Use "module spider" to find all possible modules and extensions.
Use "module keyword key1 key2 ..." to search for all possible modules matching any of the "keys".

module avail 

module avail python
Click to show 
[bbrydsoe@rackham1 ~]$ module avail python

--------------------------------- /sw/mf/rackham/applications ----------------------------------
   python_GIS_packages/3.10.8      python_ML_packages/3.9.5-gpu         wrf-python/1.3.1
   python_ML_packages/3.9.5-cpu    python_ML_packages/3.11.8-cpu (D)

----------------------------------- /sw/mf/rackham/compilers -----------------------------------
   python/2.7.6     python/3.4.3    python/3.9.5         python3/3.6.0     python3/3.11.4
   python/2.7.9     python/3.5.0    python/3.10.8        python3/3.6.8     python3/3.11.8
   python/2.7.11    python/3.6.0    python/3.11.4        python3/3.7.2     python3/3.12.1
   python/2.7.15    python/3.6.8    python/3.11.8        python3/3.8.7     python3/3.12.7 (D)
   python/3.3       python/3.7.2    python/3.12.1        python3/3.9.5
   python/3.3.1     python/3.8.7    python/3.12.7 (D)    python3/3.10.8

  Where:
   D:  Default Module

Use "module spider" to find all possible modules and extensions.
Use "module keyword key1 key2 ..." to search for all possible modules matching any of the "keys".

Ignore the comment about using module spider! 

module avail python
Click to show 
[x_birbr@tetralith3 ~]$ module avail python

--------------------- /software/sse2/tetralith_el9/modules ---------------------
   ASE/3.22.1-hpc1-python
   ASE/3.23.0-hpc1-python
   GaussSum/3.0.2-Python-3.7.8-hpc1
   NCL/6.6.2-hpc2-Python
   Pylint/3.2.5-hpc1-Python-3.11.5
   Python/recommendation                (D)
   Python/2.7.18-bare-hpc1-gcc-2022a-eb
   Python/3.10.4-bare-hpc1-gcc-2022a-eb
   Python/3.10.4-env-hpc1-gcc-2022a-eb
   Python/3.10.4-env-hpc2-gcc-2022a-eb
   Python/3.11.5-bare-hpc1-gcc-2023b-eb
   Python/3.11.5-env-hpc1-gcc-2023b-eb

  Where:
   D:  Default Module

Exercise

  1. At your chosen centre, check the different output for module avail Python and module spider Python
  2. Do you get the same output for “Python” and “python”?

List loaded modules

There is a very useful command to list which modules you have loaded.

It is

module list

or, in short form

ml

Tip

Try it now!

With nothing loaded, only the “sticky” modules are listed. They are modules that are needed for the environment to function correctly, so do not remove them!

Load software module

In order to load modules, we need to know how. This differs by center only inasmuch as some centers have prerequisites (usually compiler toolchains) for most of the software modules.

The centers that recommend module spider for finding the software modules generally have prerequisites (HPC2N, LUNARC, PDC - and sometimes C3SE), while those who recommend module avail generally do not (UPPMAX, NSC).

Thus; if the centers suggests you do module spider you will need to check the required prerequisites to load.

Prerequisites

  • This is done with module spider <module>/<version>
  • You can also do this at the centers recommending module avail, but that will generally just tell you the module can be loaded directly.

Prerequisites

At some centres it is common that modules have prerequisites (usually GCC, various libraries or something similar) while at other centres most modules can be loaded directly.

  • HPC2N: most software modules have prerequisites (GCC/Intel, OpenMPI, …), a few do not (MATLAB, …)
  • LUNARC: most software modules have prerequisites (GCC/Intel, OpenMPI, …), a few do not (MATLAB, …)
  • UPPMAX: software modules can generally be loaded directly
  • NSC: software modules can generally be loaded directly
  • PDC: most software modules have prerequisites (PDC which is a collection of compilers and libraries), a few do not (MATLAB, cray-modules, …)
  • C3SE: software modules can generally be loaded directly

Example, HPC2N

Finding out how to load Python, version 3.12.3 

b-an01 [~]$ module spider Python/3.12.3

--------------------------------------------------------------------------------------------------------------------
  Python: Python/3.12.3
--------------------------------------------------------------------------------------------------------------------
    Description:
      Python is a programming language that lets you work more quickly and integrate your systems more effectively.


    You will need to load all module(s) on any one of the lines below before the "Python/3.12.3" module is available to load.

      GCCcore/13.3.0

    This module provides the following extensions:

       flit_core/3.9.0 (E), packaging/24.0 (E), pip/24.0 (E), setuptools/70.0.0 (E), setuptools_scm/8.1.0 (E), tomli/2.0.1 (E), typing_extensions/4.11.0 (E), wheel/0.43.0 (E)

    Help:
      Description
      ===========
      Python is a programming language that lets you work more quickly and integrate your systems more effectively.


      More information
      ================
       - Homepage: https://python.org/


      Included extensions
      ===================
      flit_core-3.9.0, packaging-24.0, pip-24.0, setuptools-70.0.0, setuptools_scm-8.1.0, tomli-2.0.1, typing_extensions-4.11.0, wheel-0.43.0

There are some things to pay attention to here:

  • You are told the prerequisites are “GCCcore/13.3.0”
  • After loading that, you can load the module itself, “Python/3.12.3”
  • You are told about the extensions that are included. Here, this would be the Python packages that are included, which is not very many. HPC2N and some of the other centres use separate modules or module bundles (SciPy-bundle for instance) for any extensions/packages that you might need.

Example, UPPMAX

Finding out how to load Python, version 3.11.8 

[bbrydsoe@rackham1 ~]$ ml avail python/3.11.8

----------------------------------- /sw/mf/rackham/compilers -----------------------------------
   python/3.11.8

Use "module spider" to find all possible modules and extensions.
Use "module keyword key1 key2 ..." to search for all possible modules matching any of the "keys".

This did not tell us much! Let us try module spider 

[bbrydsoe@rackham1 ~]$ module spider python/3.11.8

--------------------------------------------------------------------------------------------
  python: python/3.11.8
--------------------------------------------------------------------------------------------

    This module can be loaded directly: module load python/3.11.8

    Help:

      Python - use python/3.11.8

          Version 3.11.8

          https://python.org

      This module was built with 

          gcc/12.3.0 sqlite/3.34.0 Tcl-Tk/8.6.11


      This module provides the executable names 'python' and 'python3'.

      Several additional python packages are also installed in this module. The complete list of packages in this module, produced using 'pip list', is:

      Package                   Version
      ------------------------- ---------------
      anndata                   0.10.5.post1
      anyio                     4.2.0
      argon2-cffi               23.1.0
      argon2-cffi-bindings      21.2.0
      array_api_compat          1.4.1
      arrow                     1.3.0
      asteval                   0.9.31
      asttokens                 2.4.1
      async-lru                 2.0.4
      ...

Here we learn:

  • You can load “python/3.11.8” directly
  • It was built with “gcc/12.3.0 sqlite/3.34.0 Tcl-Tk/8.6.11” (might be useful for compatibility with other things).
  • It lists all the extensions (here Python packages) loaded with it - and it is a lot, around 250 packages.

Loading

To load a software module, do:

  • (if needed) module load <prerequisite>/<suitable version>
  • module load <module>/<compatible version>

and the versions you got from module spider.

Again, ml load <module>/<version> can be used as a short form.

When you have loaded the module, you can see that your list of loaded modules has changed. This is done with module list or ml.

Hint

Type along!

Example, HPC2N

Loading Python 3.12.3 and prerequisites, and checking before and after which modules are loaded.

module list
Click to show output! 
b-an01 [~]$ module list

Currently Loaded Modules:
  1) snicenvironment (S)   2) systemdefault (S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge

module load GCCcore/13.3.0 Python/3.12.3

or load them on separate lines 

module load GCCcore/13.3.0
module load Python/3.12.3

What is the advantage to loading them one at a time? You can then easier find compatible modules that depend on that version, using module avail.

Click to show output! 
b-an01 [~]$ module load GCCcore/13.3.0 Python/3.12.3
b-an01 [~]$

And now do module list again!

Click to show output! 
b-an01 [~]$ module list

Currently Loaded Modules:
  1) snicenvironment (S)   4) zlib/1.3.1      7) ncurses/6.5      10) SQLite/3.45.3  13) OpenSSL/3
  2) systemdefault   (S)   5) binutils/2.42   8) libreadline/8.2  11) XZ/5.4.5       14) Python/3.12.3
  3) GCCcore/13.3.0        6) bzip2/1.0.8     9) Tcl/8.6.14       12) libffi/3.4.5

  Where:
   S:  Module is Sticky, requires --force to unload or purge

Example, NSC

Loading Python version 3.11.5 (the module is called Python/3.11.5-env-hpc1-gcc-2023b-eb and can be loaded directly). Also do module list before and after! 

module list
Click to show output! 
[x_birbr@tetralith3 ~]$ module list

Currently Loaded Modules:
  1) hpc/.1.10.1 (H,S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge
   H:             Hidden Module

module load Python/3.11.5-env-hpc1-gcc-2023b-eb
Click to show output! 
[x_birbr@tetralith3 ~]$ module load Python/3.11.5-env-hpc1-gcc-2023b-eb
[x_birbr@tetralith3 ~]$ 

ml
Click to show output! 
[x_birbr@tetralith3 ~]$ ml

Currently Loaded Modules:
  1) hpc/.1.10.1 (H,S)   2) Python/3.11.5-env-hpc1-gcc-2023b-eb

  Where:
   S:  Module is Sticky, requires --force to unload or purge
   H:             Hidden Module

Important

  • You can do several module load on the same line. Or you can do them one at a time, as you want.
    • The modules have to be loaded in order! You cannot list the prerequisite after the module that needs it!
  • One advantage to loading modules one at a time is that you can then find compatible modules that depend on that version easily.
    • Example: you have loaded GCC/13.2.0 and Python/3.11.5 at HPC2N or LUNARC. You can now do ml av or module avail to see which versions of other modules you want to load, say SciPy-bundle, are compatible.
    • If you know the name of the module you want, you can even start writing module load SciPy-bundle/ (for instance) and press TAB - the system will then autocomplete to the compatible one(s).

Loading several modules

There are situations where you need to load several modules, even at centres where there are no prerequisites.

Example

You need some Python packages not included in the base Python you loaded, for instance TensorFlow or maybe AlphaFold.

You now need to load a version of that module which is compatible with the Python you want.

Tensorflow needs Python and SciPy-bundle. SciPy-bundle also needs OpenMPI. TensorFlow loads Python and SciPy-bundle itself, but you need their prerequisites: 

b-an01 [~]$ module spider TensorFlow/2.15.1-CUDA-12.1.1

--------------------------------------------------------------------------------------------------------------------
  TensorFlow: TensorFlow/2.15.1-CUDA-12.1.1
--------------------------------------------------------------------------------------------------------------------
    Description:
      An open-source software library for Machine Intelligence


    You will need to load all module(s) on any one of the lines below before the "TensorFlow/2.15.1-CUDA-12.1.1" module is available to load.

      GCC/12.3.0  OpenMPI/4.1.5

    This module provides the following extensions:

       absl-py/2.1.0 (E), astor/0.8.1 (E), astunparse/1.6.3 (E), cachetools/5.3.3 (E), google-auth-oauthlib/1.2.0 (E), google-auth/2.29.0 (E), google-pasta/0.2.0 (E), gviz-api/1.10.0 (E), keras/2.15.0 (E), Markdown/3.6 (E), oauthlib/3.2.2 (E), pyasn1-modules/0.4.0 (E), requests-oauthlib/2.0.0 (E), rsa/4.9 (E), tblib/3.0.0 (E), tensorboard-data-server/0.7.2 (E), tensorboard-plugin-profile/2.15.1 (E), tensorboard/2.15.2 (E), tensorflow-estimator/2.15.0 (E), TensorFlow/2.15.1 (E), termcolor/2.3.0 (E), Werkzeug/3.0.2 (E), wrapt/1.14.1 (E)

b-an01 [~]$ module load GCC/12.3.0 OpenMPI/4.1.5
b-an01 [~]$ module load TensorFlow/2.15.1-CUDA-12.1.1

TensorFlow is included in the Python_ML_packages. Python gets loaded with the Python_ML_packages so nothing extra needs loading here 

[bbrydsoe@rackham1 ~]$ ml spider python_ML_packages/3.11.8-cpu

--------------------------------------------------------------------------------------------
  python_ML_packages: python_ML_packages/3.11.8-cpu
--------------------------------------------------------------------------------------------

    This module can be loaded directly: module load python_ML_packages/3.11.8-cpu

    Help:
      python_ML_packages version 3.11.8-cpu

[bbrydsoe@rackham1 ~]$ ml python_ML_packages/3.11.8-cpu
[bbrydsoe@rackham1 ~]$

However, if you for instance need Matlab together with Gurobi, then you need to load both modules: 

[bbrydsoe@rackham1 ~]$ ml spider matlab/R2023b
[bbrydsoe@rackham1 ~]$ ml Gurobi/11.0.3

Unload software modules

Aside from module unload this section will also cover module purge!

Note

Why would you want to unload a module?

  • If you want to use a different version of a module, you need to unload the current one first
  • If you need a different version of a dependent module, but it is not compatible with the current prerequisite you have loaded

Modules can be unloaded with:

  • module unload <MODULE> may or may not work
  • module unload <MODULE>/<version>
  • ml unload <MODULE>/<version> short form of the above
  • module -<MODULE>/<version>
  • ml -<MODULE>/<version> short form of the above
  • ml -<MODULE>/<version> short form of the above

Unloading a module will not unload the prerequisites.

This is not usually a problem at the centres where there rarely are prerequisites (UPPMAX, NSC), but can be annoying at centres who usually have prerequisites for the modules (HPC2N, LUNARC, C3SE, PDC). However, there the easy solution is module purge (HPC2N, LUNARC, C3SE).

The command module purge removes all the loaded modules, except the “sticky” modules. It is recommended at HPC2N, LUNARC, and C3SE.

Warning

Do not use module purge at PDC!

At PDC, there are a lot of necessary, “system-modules” that are preloaded. When you do module purge they will also be unloaded and things may not work as it should!

At UPPMAX, the system-module “uppmax” will get unloaded with module purge, but can easily be reloaded with module load uppmax.

  • To see what the uppmax module does, do: ml show uppmax. It sets some environment variables and aliases.

Unloading examples

No prerequisites

Unloading one module, with no prerequisites (for clarity, we also do module list before and after to show what is happening.

Tip

Type along!

First load a suitable module for your center (with no prerequisites). Suggestions:

  • HPC2N, LUNARC: GCC/12.3.0
  • UPPMAX: python/3.11.8
  • C3SE: Python/3.12.3-GCCcore-13.3.0
  • NSC: Python/3.11.5-env-hpc1-gcc-2023b-eb
  • PDC: cray-python/3.11.5

HPC2N

Check which modules are loaded (after loading GCC/12.3.0 earlier)

b-an01 [~]$ module list

Currently Loaded Modules:
  1) snicenvironment (S)   3) GCCcore/12.3.0   5) binutils/2.40
  2) systemdefault   (S)   4) zlib/1.2.13      6) GCC/12.3.0

  Where:
   S:  Module is Sticky, requires --force to unload or purge
b-an01 [~]$

Unload GCC/12.3.0 

b-an01 [~]$ module unload GCC/12.3.0
b-an01 [~]$

Check which modules are loaded 

b-an01 [~]$ module list

Currently Loaded Modules:
  1) snicenvironment (S)   2) systemdefault (S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge

NSC

Check which modules are loaded (after loading Python/3.11.5-env-hpc1-gcc-2023b-eb earlier) 

[x_birbr@tetralith3 ~]$ module list    
Currently Loaded Modules:
  1) hpc/.1.10.1 (H,S)   2) Python/3.11.5-env-hpc1-gcc-2023b-eb

  Where:
   S:  Module is Sticky, requires --force to unload or purge
   H:             Hidden Module

Unload Python/3.11.5-env-hpc1-gcc-2023b-eb 

[x_birbr@tetralith3 ~]$ ml -Python/3.11.5-env-hpc1-gcc-2023b-eb
[x_birbr@tetralith3 ~]$

Check which modules are loaded 

[x_birbr@tetralith3 ~]$ ml

Currently Loaded Modules:
  1) hpc/.1.10.1 (H,S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge
   H:             Hidden Module

Prerequisites

Here we look at what happens when you unload something that has a prerequisite.

Example, HPC2N

LUNARC would be the same.

Loading Python and prerequisites

b-an01 [~]$ module load GCC/12.3.0
b-an01 [~]$ module load Python/3.11.3

Check loaded modules

b-an01 [~]$ ml

Currently Loaded Modules:
  1) snicenvironment (S)   4) zlib/1.2.13     7) bzip2/1.0.8      10) Tcl/8.6.13     13) libffi/3.4.4
  2) systemdefault   (S)   5) binutils/2.40   8) ncurses/6.4      11) SQLite/3.42.0  14) OpenSSL/1.1
  3) GCCcore/12.3.0        6) GCC/12.3.0      9) libreadline/8.2  12) XZ/5.4.2       15) Python/3.11.3

  Where:
   S:  Module is Sticky, requires --force to unload or purge

Unload Python/3.11.3

b-an01 [~]$ ml unload Python/3.11.3

Check loaded modules 

b-an01 [~]$ ml

Currently Loaded Modules:
  1) snicenvironment (S)   3) GCCcore/12.3.0   5) binutils/2.40
  2) systemdefault   (S)   4) zlib/1.2.13      6) GCC/12.3.0

  Where:
   S:  Module is Sticky, requires --force to unload or purge

Unload GCC/12.3.0

b-an01 [~]$ ml -GCC/12.3.0

Check loaded modules

b-an01 [~]$ ml

Currently Loaded Modules:
  1) snicenvironment (S)   2) systemdefault (S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge

What happens if you try and unload the prerequisite?

Example, HPC2N

b-an01 [~]$ ml GCC/12.3.0 Python/3.11.3
b-an01 [~]$ ml

Currently Loaded Modules:
  1) snicenvironment (S)   4) zlib/1.2.13     7) bzip2/1.0.8      10) Tcl/8.6.13     13) libffi/3.4.4
  2) systemdefault   (S)   5) binutils/2.40   8) ncurses/6.4      11) SQLite/3.42.0  14) OpenSSL/1.1
  3) GCCcore/12.3.0        6) GCC/12.3.0      9) libreadline/8.2  12) XZ/5.4.2       15) Python/3.11.3

  Where:
   S:  Module is Sticky, requires --force to unload or purge
b-an01 [~]$ ml -GCC/12.3.0

Inactive Modules:
  1) OpenSSL/1.1       3) SQLite/3.42.0     5) XZ/5.4.2        7) libffi/3.4.4
  2) Python/3.11.3     4) Tcl/8.6.13        6) bzip2/1.0.8     8) libreadline/8.2

Due to MODULEPATH changes, the following have been reloaded:
  1) binutils/2.40     2) ncurses/6.4     3) zlib/1.2.13

module purge

What about module purge?

Important

At some centres, module purge is recommended and at some it is at least safe: HPC2N, LUNARC, NSC, C3SE 

b-an01 [~]$ ml GCC/12.3.0 Python/3.11.3
b-an01 [~]$ ml

Currently Loaded Modules:
  1) snicenvironment (S)   4) zlib/1.2.13     7) bzip2/1.0.8      10) Tcl/8.6.13     13) libffi/3.4.4
  2) systemdefault   (S)   5) binutils/2.40   8) ncurses/6.4      11) SQLite/3.42.0  14) OpenSSL/1.1
  3) GCCcore/12.3.0        6) GCC/12.3.0      9) libreadline/8.2  12) XZ/5.4.2       15) Python/3.11.3

  Where:
   S:  Module is Sticky, requires --force to unload or purge



b-an01 [~]$ ml purge
The following modules were not unloaded:
  (Use "module --force purge" to unload all):

  1) snicenvironment   2) systemdefault
b-an01 [~]$ ml

Currently Loaded Modules:
  1) snicenvironment (S)   2) systemdefault (S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge

All good! 

[bbrydsoe@cosmos3 ~]$ ml GCC/12.3.0 Python/3.11.3
[bbrydsoe@cosmos3 ~]$ ml

Currently Loaded Modules:
  1) SoftwareTree/Milan (S)   4) binutils/2.40   7) ncurses/6.4      10) SQLite/3.42.0  13) OpenSSL/1.1
  2) GCCcore/12.3.0           5) GCC/12.3.0      8) libreadline/8.2  11) XZ/5.4.2       14) Python/3.11.3
  3) zlib/1.2.13              6) bzip2/1.0.8     9) Tcl/8.6.13       12) libffi/3.4.4

  Where:
   S:  Module is Sticky, requires --force to unload or purge



[bbrydsoe@cosmos3 ~]$ ml purge
The following modules were not unloaded:
  (Use "module --force purge" to unload all):

  1) SoftwareTree/Milan
[bbrydsoe@cosmos3 ~]$ ml

Currently Loaded Modules:
  1) SoftwareTree/Milan (S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge

All good!

[brydso@alvis1 ~]$ ml
No modules loaded
[brydso@alvis1 ~]$ ml Python/3.12.3-GCCcore-13.3.0
[brydso@alvis1 ~]$ ml

Currently Loaded Modules:
  1) GCCcore/13.3.0                 5) ncurses/6.5-GCCcore-13.3.0       9) XZ/5.4.5-GCCcore-13.3.0
  2) zlib/1.3.1-GCCcore-13.3.0      6) libreadline/8.2-GCCcore-13.3.0  10) libffi/3.4.5-GCCcore-13.3.0
  3) binutils/2.42-GCCcore-13.3.0   7) Tcl/8.6.14-GCCcore-13.3.0       11) OpenSSL/3
  4) bzip2/1.0.8-GCCcore-13.3.0     8) SQLite/3.45.3-GCCcore-13.3.0    12) Python/3.12.3-GCCcore-13.3.0



[brydso@alvis1 ~]$ ml purge
[brydso@alvis1 ~]$ ml
No modules loaded

All good! No modules before that got lost! 

[x_birbr@tetralith3 ~]$ ml Python/3.11.5-env-hpc1-gcc-2023b-eb
[x_birbr@tetralith3 ~]$ ml 

Currently Loaded Modules:
  1) hpc/.1.10.1 (H,S)   2) Python/3.11.5-env-hpc1-gcc-2023b-eb

  Where:
   S:  Module is Sticky, requires --force to unload or purge
   H:             Hidden Module
[x_birbr@tetralith3 ~]$ ml purge
The following modules were not unloaded:
  (Use "module --force purge" to unload all):

  1) hpc/.1.10.1
[x_birbr@tetralith3 ~]$ ml

Currently Loaded Modules:
  1) hpc/.1.10.1 (H,S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge
   H:             Hidden Module

All good!

Important

While at other centres, it is more or less problematic to use module purge: UPPMAX, PDC 

[bbrydsoe@rackham1 ~]$ ml python/3.11.8
[bbrydsoe@rackham1 ~]$ ml

Currently Loaded Modules:
  1) uppmax   2) python/3.11.8



[bbrydsoe@rackham1 ~]$ ml purge
[bbrydsoe@rackham1 ~]$ ml
No modules loaded

Warning! You need to reload “uppmax” module! module load uppmax

bbrydsoe@login1:~> ml

Currently Loaded Modules:
  1) craype-x86-rome                        6) cce/17.0.0           11) PrgEnv-cray/8.5.0
  2) libfabric/1.20.1                       7) craype/2.7.30        12) snic-env/1.0.0
  3) craype-network-ofi                     8) cray-dsmml/0.2.2     13) systemdefault/1.0.0 (S)
  4) perftools-base/23.12.0                 9) cray-mpich/8.1.28
  5) xpmem/2.8.2-1.0_3.9__g84a27a5.shasta  10) cray-libsci/23.12.5

  Where:
   S:  Module is Sticky, requires --force to unload or purge



bbrydsoe@login1:~> ml cray-python/3.11.5
bbrydsoe@login1:~> ml

Currently Loaded Modules:
  1) craype-x86-rome                        6) cce/17.0.0           11) PrgEnv-cray/8.5.0
  2) libfabric/1.20.1                       7) craype/2.7.30        12) snic-env/1.0.0
  3) craype-network-ofi                     8) cray-dsmml/0.2.2     13) systemdefault/1.0.0 (S)
  4) perftools-base/23.12.0                 9) cray-mpich/8.1.28    14) cray-python/3.11.5
  5) xpmem/2.8.2-1.0_3.9__g84a27a5.shasta  10) cray-libsci/23.12.5

  Where:
   S:  Module is Sticky, requires --force to unload or purge



bbrydsoe@login1:~> ml purge
The following modules were not unloaded:
  (Use "module --force purge" to unload all):

  1) systemdefault/1.0.0
bbrydsoe@login1:~> ml

Currently Loaded Modules:
  1) systemdefault/1.0.0 (S)

  Where:
   S:  Module is Sticky, requires --force to unload or purge

WARNING WARNING WARNING !!! Lots of modules got unloaded! Either logout and login again, or make sure you have saved a list of the system-modules so you can reload all of them (or use module collections - which we get to soon.

Exercise

  1. Check how to load R at your centre. Pick a version.
  2. Does it have prerequisites at your centre?
  3. Load R for a specific version (but first load prerequisites if there are any).
  4. Run module list to see what modules got loaded. Was it more than you expected?
  5. Unload R (and prerequisites if there are any). See what happens.
  6. Do module list again.

module show

This command shows commands in the module file (MODULE) and can be used to list information about modules.

  • at centres with a “flat” structure this can be used on all modules since there are no prerequisites: UPPMAX, NSC
  • at centres with a “hierarchial” structure this can only be used on the modules that are currently available to load (and can be seen with module avail): HPC2N, LUNARC, C3SE, PDC

Example, HPC2N

Let us look at CUDA: 

b-an01 [~]$ ml show CUDA/12.6.0
----------------------------------------------------------------------------------------------------------------------
   /hpc2n/eb/modules/all/Core/CUDA/12.6.0.lua:
----------------------------------------------------------------------------------------------------------------------
help([[
Description
===========
CUDA (formerly Compute Unified Device Architecture) is a parallel
 computing platform and programming model created by NVIDIA and implemented by the
 graphics processing units (GPUs) that they produce. CUDA gives developers access
 to the virtual instruction set and memory of the parallel computational elements in CUDA GPUs.


More information
================
 - Homepage: https://developer.nvidia.com/cuda-toolkit
]])
whatis("Description: CUDA (formerly Compute Unified Device Architecture) is a parallel
 computing platform and programming model created by NVIDIA and implemented by the
 graphics processing units (GPUs) that they produce. CUDA gives developers access
 to the virtual instruction set and memory of the parallel computational elements in CUDA GPUs.")
whatis("Homepage: https://developer.nvidia.com/cuda-toolkit")
whatis("URL: https://developer.nvidia.com/cuda-toolkit")
conflict("CUDA")
prepend_path("CMAKE_PREFIX_PATH","/hpc2n/eb/software/CUDA/12.6.0")
prepend_path("CPATH","/hpc2n/eb/software/CUDA/12.6.0/include")
prepend_path("CPATH","/hpc2n/eb/software/CUDA/12.6.0/extras/CUPTI/include")
prepend_path("CPATH","/hpc2n/eb/software/CUDA/12.6.0/nvvm/include")
prepend_path("LD_LIBRARY_PATH","/hpc2n/eb/software/CUDA/12.6.0/lib")
prepend_path("LD_LIBRARY_PATH","/hpc2n/eb/software/CUDA/12.6.0/extras/CUPTI/lib64")
prepend_path("LD_LIBRARY_PATH","/hpc2n/eb/software/CUDA/12.6.0/nvvm/lib64")
prepend_path("LIBRARY_PATH","/hpc2n/eb/software/CUDA/12.6.0/lib")
prepend_path("LIBRARY_PATH","/hpc2n/eb/software/CUDA/12.6.0/extras/CUPTI/lib64")
prepend_path("LIBRARY_PATH","/hpc2n/eb/software/CUDA/12.6.0/nvvm/lib64")
prepend_path("LIBRARY_PATH","/hpc2n/eb/software/CUDA/12.6.0/stubs/lib64")
prepend_path("PATH","/hpc2n/eb/software/CUDA/12.6.0/bin")
prepend_path("PATH","/hpc2n/eb/software/CUDA/12.6.0/nvvm/bin")
prepend_path("PKG_CONFIG_PATH","/hpc2n/eb/software/CUDA/12.6.0/pkgconfig")
prepend_path("XDG_DATA_DIRS","/hpc2n/eb/software/CUDA/12.6.0/share")
setenv("EBROOTCUDA","/hpc2n/eb/software/CUDA/12.6.0")
setenv("EBVERSIONCUDA","12.6.0")
setenv("EBDEVELCUDA","/hpc2n/eb/software/CUDA/12.6.0/easybuild/Core-CUDA-12.6.0-easybuild-devel")
setenv("CUDA_HOME","/hpc2n/eb/software/CUDA/12.6.0")
setenv("CUDA_ROOT","/hpc2n/eb/software/CUDA/12.6.0")
setenv("CUDA_PATH","/hpc2n/eb/software/CUDA/12.6.0")

buildenv

The most useful usage of module show is with the buildenv module. If you load a “compiler toolchain” (see next section) and then the “buildenv” module, you can do 

module show buildenv

to see all the various useful environment variables that can now be accessed for linking with when you are building something.

If you want to read more about this, you can check the buildenv - build environment section under “EXTRA” -> “Advanced module commands”.

module help

This command prints the list of possible commands and can also be used to get the help message from module(s).

  • at centres with a “flat” structure this can be used on all modules since there are no prerequisites: UPPMAX, NSC
  • at centres with a “hierarchial” structure this can only be used on the modules that are currently available to load (and can be seen with module avail): HPC2N, LUNARC, C3SE, PDC

Example, HPC2N

Let us look at CUDA:

b-an01 [~]$ module help CUDA/12.6.0

----------------------------------------- Module Specific Help for "CUDA/12.6.0" -----------------------------------------

Description
===========
CUDA (formerly Compute Unified Device Architecture) is a parallel
 computing platform and programming model created by NVIDIA and implemented by the
 graphics processing units (GPUs) that they produce. CUDA gives developers access
 to the virtual instruction set and memory of the parallel computational elements in CUDA GPUs.


More information
================
 - Homepage: https://developer.nvidia.com/cuda-toolkit

module save/restore

Module collections are used to load/unload a bunch of modules: module save <collection> and module restore <collection>.

This can be useful if you often need to load the same several modules in specific versions, for instance.

Creating a module collection

  1. Load the modules you need.
  2. Save the collection (you can name it as you want, here MYMODULES): 
    module save MYMODULES

Example

Assuming we need pandas and matplotlib at HPC2N (very similar to LUNARC):

  1. Load the modules 
    b-an01 [~]$ module load GCC/12.3.0 
b-an01 [~]$ module load Python/3.11.3 
b-an01 [~odule load OpenMPI/4.1.5 
b-an01 [~]$ module load SciPy-bundle/2023.07 
b-an01 [~]$ module load matplotlib/3.7.2

  2. Save to a collection (here, “mypython”) 

    b-an01 [~]$ module save mypython
Saved current collection of modules to: "mypython"

b-an01 [~]$

  3. Then, maybe later you find you also need mpi4py so you load it and add it to the collection: 

    b-an01 [~]$ ml mpi4py/3.1.4 
b-an01 [~]$ 
module save mypython
Saved current collection of modules to: "mypython"

b-an01 [~]$

  4. After you have been logged out and in again, or maybe unloaded/purged the modules, you can then restore it again: 

    b-an01 [~]$ module restore mypython
Restoring modules from user's mypython
b-an01 [~]$

Warning

If you are working on PDC’s cluster Dardel, and are used to the command module purge then this is a good way of restoring the system-modules:

  • Immediately after logging in (before loading or unloading anything) create a collection: module save pdcsystem (name as you want)
  • Then, after you have done module purge, just do module restore pdcsystem and you are back to the correct system setup.

Workflow - module collections

  • Create a module collection
  • Work with it
  • Possibly unload all modules and load different ones to work with
  • Then later restore the module collection again and keep working
  • Possibly add more modules and save them to the collection
  • Each time you have logged out and are logging in again, you can easily restore the modules you need
    • Much safer than having it in your .bashrc since that is something easily forgotten about and then when you suddenly need to work with different modules or different versions, maybe months later, you are wondering why it is not working as it should and the reason is that you have auto-loaded things in your .bashrc!

Exercise

  • Try loading some modules
  • create a module collection
  • do module list to see what you have
  • unload the modules
  • check with module list
  • restore the module collection
  • check with module list

Hints

Note

How would you find, for instance, installed Python package modules for a specific Python version?

Important

At centres with prerequisites, you need to load, say, GCC instead of GCCcore in order to be able to access, for instance, SciPy-bundle!!! You also need to load a compatible OpenMPI in order to be sure to get all available installed Python package modules.

Example, HPC2N

(Very similar to LUNARC)

  1. First load the Python module and prerequisites, and OpenMPI
    b-an01 [~]$ module load GCC/12.3.0
b-an01 [~]$ module load Python/3.11.3
b-an01 [~]$ module load OpenMPI/4.1.5
  2. Then you can check what Python package modules are installed (scroll down a bit):
    b-an01 [~]$ module avail
...
----------------------- This is a list of module extensions "module --nx avail ..." to not show. ------------------------
                                               (E)     fontBitstreamVera                         (E)
 ADGofTest                                     (E)     fontLiberation                            (E)
 AICcmodavg                                    (E)     fontawesome                               (E)
 ALDEx2                                        (E)     fontquiver                                (E)
 ALL                                           (E)     fonttools                                 (E)
 AMAPVox                                       (E)     forcats                                   (E)
 ANCOMBC                                       (E)     foreach                                   (E)
 ATACseqQC                                     (E)     forecast                                  (E)
 AUC                                           (E)     foreign                                   (E)
 AUCell                                        (E)     formatR                                   (E)
 Aerial-Gym-Simulator                          (E)     formula.tools                             (E)
 AgiMicroRna                                   (E)     formulaic                                 (E)
 AlgDesign                                     (E)     fossil                                    (E)
 Algorithm::Dependency                         (E)     fpc                                       (E)
 Algorithm::Diff                               (E)     fpp                                       (E)
 Alien::Base                                   (E)     fracdiff                                  (E)
 Alien::Build::Plugin::Download::GitLab        (E)     fresh                                     (E)
 Alien::Libxml2                                (E)     frozenlist                                (E)
 AlphaFold                                     (E)     fs                                        (E)
 AlphaPulldown                                 (E)     fsc.export                                (E)
 ...
  3. The “module extensions” are things that are available as part of a modules. Sometimes several for each module. You can see for instance “foreach” R package is available (the list has everything available, not just Python packages). You can then do module spider foreach to see versions and then module spider foreach/1.5.2 to see where you find that extension.
  4. If you keep going further down on the list you got with “module avail” then you see that “mpi4py” is available and that “pandas” is available. Let us see how to load these.
  5. Doing 
    b-an01 [~]$ module avail mpi4py
    tells us it can be loaded directly
  6. Doing 
    b-an01 [~]$ module avail pandas
    gives some conflicting answer, so we try with module spider pandas and then module spider pandas/version to learn that it is part of SciPy-bundle.
  7. We can then load SciPy-bundle directly and get “pandas” available.

Note

Of course, you can always try doing module spider <software or package> and module avail <software or package> directly to see if you are lucky and get the name it is called.

Summary

Note

  • We learned about the following commands:
    • load
    • unload
    • purge
    • show
    • help
    • save/restore