Strong scaling of general-purpose molecular dynamics simulations on GPUs

By Jens Glaser, T. D. Nguyen, Joshua A. Anderson, Pak Lui, F. Spiga, Jaime A. Millan, D. Morse, Sharon C. Glotzer. Computer Physics Communications, 2014

611 citations recorded by the source corpus.

Abstract We describe a highly optimized implementation of MPI domain decomposition in a GPU-enabled, general-purpose molecular dynamics code, HOOMD-blue (Anderson and Glotzer, 2013). Our approach is inspired by a traditional CPU-based code, LAMMPS (Plimpton, 1995), but is implemented within a code that was designed for execution on GPUs from the start (Anderson et al., 2008). The software supports short-ranged pair force and bond force fields and achieves optimal GPU performance using an autotuning algorithm. We are able to demonstrate equivalent or superior scaling on up to 3375 GPUs in Lennard-Jones and dissipative particle dynamics (DPD) simulations of up to 108 million particles. GPUDirect RDMA capabilities in recent GPU generations provide better performance in full double precision calculations. For a representative polymer physics application, HOOMD-blue 1.0 provides an effective GPU vs. CPU node speed-up of 12.5 × .

Subfields: Biomedical Engineering, Condensed Matter Physics, Hardware and Architecture

View Paper: https://www.semanticscholar.org/paper/f1d346ee291314ee2ab80c1d6930f5341a97264c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Condensed Matter Physics

In Hardware and Architecture

Citations in the corpus

Cohesin loss eliminates all loop domains (2017, methodology): “In the HOOMD-blue molecular dynamics package (Glaser et al., 2015) (Anderson et al.”
JAX, M.D.: End-to-End Differentiable, Hardware Accelerated, Molecular Dynamics in Pure Python (2019, methodology): “General MD packages have been widely used to simulate molecules and solids, either using firstprinciples potentials (using software packages that derive the potential from quantum mechanics (Car & Parrinello, 1985) e.g. Quantum Espresso (Giannozzi et al., 2009), VASP (Hafner, 2008), SIESTA (Soler et”
End-to-End Differentiable , Hardware Accelerated , Molecular Dynamics in Pure Python (2019, methodology): “General MD packages have been widely used to simulate molecules and solids, either using firstprinciples potentials (using software packages that derive the potential from quantum mechanics (Car & Parrinello, 1985) e.g. Quantum Espresso (Giannozzi et al., 2009), VASP (Hafner, 2008), SIESTA (Soler et”
HOOMD-blue: A Python package for high-performance molecular dynamics and hard particle Monte Carlo simulations (2013, methodology): “Typical research-relevant MD simulations execute an order of magnitude faster on a single GPU than on all cores of a single CPU socket for system sizes of four thousand particles per GPU or more [23].”