AGILE

This is the development version of AGILE, a GPU enabled fork of MPI-AMRVAC. All documentation is available on amrvac.org.

Work in progress...

To install this version, follow these steps:

• install the uv package manager pip install uv
• make sure $AMRVAC_DIR points to the repository root folder
• install the required python packages uv sync and activate them source .venv/bin/activate
• go into a test, e.g. cd tests/hd/KH3D
• to compile, load the appropriate modules, e.g. on snellius:

module purge
module load 2023
module load OpenMPI/4.1.5-NVHPC-24.5-CUDA-12.1.1

• compile with nvfortran and activated OPENACC via make arch=nvidia OPENACC=1

Currently supported features on master

• Physics modules: hd, mhd [glm], ffhd
• Source terms (gravity, radiative cooling, hyperbolic thermal conduction, user defined) and boundary conditions (symm, asymm, cont etc. but also special)
• Multi-block uniform grid and static mesh refinement (SMR) simulations
• Multi-GPU (MPI) simulations
• Adaptive mesh refinement simulations (single GPU case so far)