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, but since GROMACS is extremely fast at calculating the nonbonded interactions (that usually dominate simulations) many groups are also using it for research on non-biological systems, e.g. polymers.
You must load the appropriate modules in your submission script.
- HPC_GROMACS_DIR - installation directory
- HPC_GROMACS_BIN - executable directory
- HPC_GROMACS_INC - header file directory
- HPC_GROMACS_LIB - library directory
Gromacs 4.6.5 was built with CUDA-based GPU acceleration in both the serial and parallel (MPI+OpenMP) executables. It is not necessary to load the associated CUDA module. It will be loaded by the Gromacs 4.6.5 module. If no GPUs are available, Gromacs 4.6.5 will note that no GPUs were found and will run without them.
Also note that the use of "pbsgpu-wrapper" with gromacs 4.6.5 when attempting to use more than a single GPU will not work and Gromacs will only see the GPU exposed to the MPI "rank 0" process. Therefore, if you intend to use more than one GPU with Gromacs you should observe the following guidelines.
- Run some preliminary scaling test to ensure that you can efficiently utilize more than one GPU.
- Limit your runs to multiple GPUs within the same server (i.e. nodes=1:ppn=4:gpus=2)
- Expose the assigned GPUs to gromacs via the "CUDA_VISIBLE_GPUS" environment variable (see sample scripts).
PBS Script Examples
See the Gromacs_PBS page for Gromacs PBS script examples.
If you publish research that uses Gromacs you have to cite it as follows:
GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation Hess, B., Kutzner, C., van der Spoel, D. and Lindahl, E. J. Chem. Theory Comput., 4, 435-447 (2008)