LAMMPS-GPU - NCI National Facility
National Computational Infrastructure
NCI National Facility
Software at a site Software by category Facilities at a site Login
        Generic (all NCI systems)       Version specific (all NCI systems)       Host and version specific

What is LAMMPS-GPU LAMMPS stands for Large-scale Atomic/Molecular Massively Parallel Simulator. LAMMPS has potentials for soft materials (biomolecules, polymers) and solid-state materials (metals, semiconductors) and coarse-grain systems. It can be used to model atoms or, more generically, as a parallel particle simulator at the mesoscale or continuum levels. For more information see the LAMMPS GPU.
What is new in this Version This LAMMPS version is implemented to utilize GPU and CUDA functionalities.
How to Use LAMMPS-GPU on xe To use this version of LAMMPS set the environment module load lammps/gpu To understand more about modules see our module help document. A sample batch job script is shown as follows:
#PBS -P a99
#PBS -l walltime=10:00
#PBS -l vmem=800mb
#PBS -l ncpus=1
#PBS -l ngpus=2
#PBS -wd

module load lammps/gpu

lmp_gpu_SPDP < input

Three precision options are available for GPU package:
* lmp_gpu_SPSP : single precision only
* lmp_gpu_SPDP : mixed single and double precision
* lmp_gpu_DPDP : full double precision

GPU package provides:
* lj/cut/gpu
* lj/light/gpu
* gayberne/gpu
* lj/charmm/coul/long/gpu

Two precision options are available for USER-CUDA package:
* lmp_cuda_SPSP: single precision only
* lmp_cuda_SPDP: positions in double precision
* lmp_cuda_DPDP: full double precision

USER-CUDA package provides:
* 20 pair forces
* long range coulomb with pppm/cuda
* nve/cuda, nvt/cuda, npt/cuda
* several more important fixes and computes

To be noted: "accelerator cuda" is no longer needed in the input files to use USER-CUDA package.

Software Userguide Details please refer to, and
 Metadata for using LAMMPS-GPU on xe
 XML Query for LAMMPS-GPU on xe
Partner contact email: problems, suggestions, questions to