GROMACS on Aurora¶
What is GROMACS?¶
GROMACS is a versatile package to perform molecular dynamics, i.e. simulate the Newtonian equations of motion for systems with hundreds to millions of particles. It is primarily designed for biochemical molecules like proteins, lipids and nucleic acids that have a lot of complicated bonded interactions. However, since GROMACS is extremely fast at calculating the nonbonded interactions (which usually dominate simulations), many groups also use it for research on non-biological systems, e.g. polymers.
Using GROMACS at ALCF¶
ALCF offers assistance with building binaries and compiling instructions for GROMACS. For questions, contact us at support@alcf.anl.gov.
Building GROMACS¶
-
Download the source code. See GROMACS documentation for the latest release.
Where
https://ftp.gromacs.org/gromacs/gromacs-2026.4.tar.gzis the URL copied from the download page. -
Extract the tarball:
-
Load required modules (verify versions on Aurora with
module avail): -
Create a build directory:
-
Configure the build:
cmake .. \ -DCMAKE_C_COMPILER=mpicc \ -DCMAKE_CXX_COMPILER=mpicxx \ -DCMAKE_INSTALL_PREFIX=$HOME/software/gromacs-2026.4 \ -DGMX_MPI=ON -DGMX_OPENMP=ON -DGMX_GPU=SYCL -DGMX_SYCL=DPCPP \ -DGMX_GPU_NB_CLUSTER_SIZE=8 \ -DGMX_GPU_NB_NUM_CLUSTER_PER_CELL_X=1Where
$HOME/software/gromacs-2026.4is the install directory on Aurora. -
Compile the code:
-
Install the binaries:
The installed binary is
$HOME/software/gromacs-2026.4/bin/gmx_mpi.
Running GROMACS on Aurora¶
A sample PBS script for Aurora follows. Set <project> to your allocation and <path> to your input/output paths as described below.
Where <project> is your allocation name, home:flare is the <path> filesystem request (see qstat -Bf for valid_filesystems), and each <path> in EXE_ARG is the TPR input, checkpoint file, and output prefix respectively.
We strongly suggest that users try combinations of different numbers of nodes, MPI ranks per node, number of GPU tasks/devices, GPU task decomposition between nonbonded and PME kernels, and OMP threads per rank to find the optimal throughput for their particular workload.