• Corpus ID: 10405383

Supercomputing for Molecular Dynamics Simulations: Handling Multi-Trillion Particles in Nanofluidics

  title={Supercomputing for Molecular Dynamics Simulations: Handling Multi-Trillion Particles in Nanofluidics},
  author={Alexander Heinecke and Wolfgang Eckhardt and Martin Thomas Horsch and Hans-Joachim Bungartz},
This work presents modern implementations of relevant molecular dynamics algorithms using ls1 mardyn, a simulation program for engineering applications. The text focuses strictly on HPC-related aspects, covering implementation on HPC architectures, taking Intel Xeon and Intel Xeon Phi clusters as representatives of current platforms. The work describes distributed and shared-memory parallelization on these platforms, including load balancing, with a particular focus on the efficient… 
Optimized Force Calculation in Molecular Dynamics Simulations for the Intel Xeon Phi
The novel shared-memory parallelization scheme allows to retain Newton’s third law optimization and exhibits very good scaling on many-core devices such as a full Xeon Phi card running 240 threads.
The Vectorization of the Tersoff Multi-body Potential: An Exercise in Performance Portability
This work develops a vectorization scheme applicable to both CPUs and accelerators for multi-body potentials within the molecular dynamics code LAMMPS on several architectures, demonstrating efficiency gains not only for computational kernels, but also for large-scale simulations.
Load Balancing for Molecular Dynamics Simulations on Heterogeneous Architectures
A k-d tree decomposition method to balance load on heterogeneous compute clusters is applied and is incorporated into the molecular dynamics simulation program ls1 mardyn, which presents performance results for simulations executed on hybrid AMD Bulldozer–Intel Sandy Bridge, Intel Westmere– Intel Sandy Bridge and Intel Xeon–Intel Xeon Phi-architectures.
Coupling of particle simulation and lattice Boltzmann background flow on adaptive grids
An approach for the domain decomposition of the linked cells algorithm based on space-filling curves and the P4EST library is presented and strategies to distribute differently refined grids on parallel processes are explained, including a parallel algorithm to construct the finest common tree using P4 EST.
Energy-conserving coarse-graining of complex molecules.
The proposed non-isothermal coarse-grained model is thermodynamically consistent and opens up a first principles CG strategy for the study of energy transport issues that are not accessible using current isothermal models.
Alien Mindscapes—A Perspective on the Search for Extraterrestrial Intelligence
This approach will augment the odds of detecting a signal by broadening the understanding of the evolutionary and systemic components in the search for extraterrestrial intelligence (ETI), provide more targets for radio and optical SETI, and identify new ways of decoding and coding messages using universal markers.
Learning and Optimization for Mixed Autonomy Systems - A Mobility Context
It is demonstrated that deep reinforcement learning (RL) can serve as a unifying framework for studying the behavior of disparate and complex scenarios common in mixed autonomy systems, and that deep RL has the potential to scale to high-dimensional control systems, such as traffic networks and other mixed autonomous systems.
Efficacy of the radial pair potential approximation for molecular dynamics simulations of dense plasmas
Macroscopic simulations of dense plasmas rely on detailed microscopic information that can be computationally expensive and is difficult to verify experimentally. In this work, we delineate the


Multibillion-atom molecular dynamics simulation: Design considerations for vector-parallel processing
591 TFLOPS Multi-trillion Particles Simulation on SuperMUC
This work conducts performance and scalability studies of an optimized version of the code ls1 mardyn, and presents the implementation requiring only 32 Bytes per molecule, which allows for the, to the authors' knowledge, largest MD simulation to date.
ls1 mardyn: The massively parallel molecular dynamics code for large systems
The molecular dynamics simulation code ls1 mardyn is presented. It is a highly scalable code, optimized for massively parallel execution on supercomputing architectures and currently holds the world
An efficient vectorization of linked-cell particle simulations
An implementation of the linked-cell algorithm for the SSE and AVX instruction set, which achieves the theoretical limit for SSE, will allow the efficient usage of future architectures with wider vector units.
By utilizing the molecular dynamics code SPaSM on Livermore's BlueGene/L architecture, consisting of 212 992 IBM PowerPC440 700 MHz processors, a molecular dynamics simulation was run with one
Hardware accelerator for molecular dynamics: MDGRAPE-2
Numerical Simulation in Molecular Dynamics: Numerics, Algorithms, Parallelization, Applications
This book presents in detail both the necessary numerical methods and techniques (linked-cell method, SPME-method, tree codes, multipole technique) and the theoretical background and foundations and illustrates the aspects modelling, discretization, algorithms and their parallel implementation with MPI on computer systems with distributed memory.
Enhanced molecular dynamics performance with a programmable graphics processor
Memory-Efficient Implementation of a Rigid-Body Molecular Dynamics Simulation
  • W. Eckhardt, T. Neckel
  • Computer Science
    2012 11th International Symposium on Parallel and Distributed Computing
  • 2012
This paper investigates two distinct implementational aspects of the frequently used Linked-Cell algorithm for rigid-body molecular dynamics simulations: the representation of particle data for the force calculation, and the layout of data structures in memory.