The LXPLUS cluster is located at CERN.


If you are new to this system, please see the following resources:

  • Lxplus documentation

  • Batch system: HTCondor

  • Filesystem locations:
    • User folder: /afs/cern.ch/user/<a>/<account> (10GByte)

    • Work folder: /afs/cern.ch/work/<a>/<account> (100GByte)

    • Eos storage: /eos/home-<a>/<account> (1T)

Through LXPLUS we have access to CPU and GPU nodes (the latter equipped with NVIDIA V100 and T4 GPUs).


Only very little software is pre-installed on LXPLUS so we show how to install from scratch all the dependencies using Spack.

For size reasons it is not advisable to install WarpX in the $HOME directory, while it should be installed in the “work directory”. For this purpose we set an environment variable with the path to the “work directory”

export WORK=/afs/cern.ch/work/${USER:0:1}/$USER/

We clone WarpX in $WORK:

cd $WORK
git clone https://github.com/ECP-WarpX/WarpX.git warpx

Installation profile file

The easiest way to install the dependencies is to use the pre-prepared warpx.profile as follows:

cp $WORK/warpx/WarpX/Tools/machines/lxplus-cern/lxplus_warpx.profile.example $WORK/lxplus_warpx.profile
source $WORK/lxplus_warpx.profile

When doing this one can directly skip to the Building WarpX section.

To have the environment activated at every login it is then possible to add the following lines to the .bashrc

export WORK=/afs/cern.ch/work/${USER:0:1}/$USER/
source $WORK/lxplus_warpx.profile


The pre-installed GNU compiler is outdated so we need a more recent compiler. Here we use the gcc 11.2.0 from the LCG project, but other options are possible.

We activate it by doing

source /cvmfs/sft.cern.ch/lcg/releases/gcc/11.2.0-ad950/x86_64-centos7/setup.sh

In order to avoid using different compilers this line could be added directly into the $HOME/.bashrc file.


We download and activate Spack in $WORK:

cd $WORK
git clone -c feature.manyFiles=true https://github.com/spack/spack.git
source spack/share/spack/setup-env.sh

Now we add our gcc 11.2.0 compiler to spack:

spack compiler find /cvmfs/sft.cern.ch/lcg/releases/gcc/11.2.0-ad950/x86_64-centos7/bin

Installing the Dependencies

To install the dependencies we create a virtual environment, which we call warpx-lxplus:

spack env create warpx-lxplus $WORK/WarpX/Tools/machines/lxplus-cern/spack.yaml
spack env activate warpx-lxplus
spack install

If the GPU support or the Python bindings are not needed, it’s possible to skip the installation by respectively setting the following environment variables export SPACK_STACK_USE_PYTHON=0 and export SPACK_STACK_USE_CUDA = 0 before running the previous commands.

After the installation is done once, all we need to do in future sessions is just activate the environment again:

spack env activate warpx-lxplus

The environment warpx-lxplus (or -cuda or -cuda-py) must be reactivated everytime that we log in so it could be a good idea to add the following lines to the .bashrc:

source $WORK/spack/share/spack/setup-env.sh
spack env activate -d warpx-lxplus
cd $HOME

Building WarpX

We prepare and load the Spack software environment as above. Then we build WarpX:

cmake -S . -B build -DWarpX_DIMS="1;2;RZ;3"
cmake --build build -j 6

Or if we need to compile with CUDA:

cmake -S . -B build -DWarpX_COMPUTE=CUDA -DWarpX_DIMS="1;2;RZ;3"
cmake --build build -j 6

That’s it! A 3D WarpX executable is now in build/bin/ and can be run with a 3D example inputs file. Most people execute the binary directly or copy it out to a location in $WORK.

Python Bindings

Here we assume that a Python interpreter has been set up as explained previously.

Now, ensure Python tooling is up-to-date:

python3 -m pip install -U pip
python3 -m pip install -U build packaging setuptools wheel
python3 -m pip install -U cmake

Then we compile WarpX as in the previous section (with or without CUDA) adding -DWarpX_PYTHON=ON and then we install it into our Python:

cmake -S . -B build -DWarpX_COMPUTE=CUDA -DWarpX_DIMS="1;2;RZ;3" -DWarpX_APP=OFF -DWarpX_PYTHON=ON
cmake --build build --target pip_install -j 6

This builds WarpX for 3D geometry.

Alternatively, if you like to build WarpX for all geometries at once, use:

BUILD_PARALLEL=6 python3 -m pip wheel .
python3 -m pip install pywarpx-*whl