Software

Software#

Operating system#

The KTA Computer System runs on AlmaLinux 9, a Linux distribution which is binary compatible with Red Hat Enterprise Linux (RHEL). It also officially supported by CERN.

System-wide available software#

The software repositories of Almalinux contain a wide variety of scientific software. This software is available system-wide and can be used directly without any additional setup. Here is a small non-exhaustive list of available packages:

  • gcc

  • clang

  • python

  • ROOT

  • git

  • make

  • cmake

  • and many more

In case the software you need for your workflow or analysis is missing, you can request its installation by contacting ktas.admin@ph.tum.de. If the software cannot be installed from the repositories, find below additional methods to install software locally. You can of course also compile the software you need by yourself.

Anaconda#

Anaconda is a well-known distribution of the Python language for scientific computing which comes equipped with a package manager and environment management system known as conda. Users can install anaconda by themselves, but we offer a global installation of anaconda on all machines in our cluster. You can enter the conda environment by executing this command

[ga45can@nidoking ~]$ source /opt/anaconda3/bin/activate
(base) [ga45can@nidoking ~]$

which will enter you in the base environment. Note that the base environment cannot be changed by users. But from here you can create your very own environment and install the software you need. Here is an example

(base) [ga45can@nidoking ~]$ conda activate MyAwesomeDevEnv
(...terminal output...)
(MyAwesomeDevEnv) [ga45can@nidoking ~]$ conda install tensorflow root
(...terminal output...)
(MyAwesomeDevEnv) [ga45can@nidoking ~]$ which root
/MyAwesomeDevEnv/bin/root
(MyAwesomeDevEnv) [ga45can@nidoking ~]$ python -c "import tensorflow as tf; print(tf.__version__)"
2.6.2

We recommend to configure conda by creating a $HOME/.condarc file and using these settings

channels:
  - conda-forge
  - defaults
channel_priority: strict
envs_dirs:
  - /Path/not/in/your/home/directory

this will activate the conda-forge repository which unlocks more software. Additionally, conda environments can get very large and space in the home directories is limited. With env_dirs users can define the software to be stored outside off the home directory. This is strongly recommended.

Spack#

Spack is a package manager used in HPC environments. User can activate spack with this command.

[ga45can@nidoking ~]$ source /tank1/spack/spack/share/spack/setup-env.sh

In contrast to anaconda as described above, users cannot install packages by themselves because the packages cache is shared by all users. All available packages by spack can be searched here. Users must contact ktas.admins@ph.tum.de to request installation of packages.