Optimizing legacy molecular dynamics software with directive-based offload
Abstract
The directive-based programming models are one solution for exploiting many-core coprocessors to increase simulation rates in molecular dynamics. They offer the potential to reduce code complexity with offload models that can selectively target computations to run on the CPU, the coprocessor, or both. In our paper, we describe modifications to the LAMMPS molecular dynamics code to enable concurrent calculations on a CPU and coprocessor. We also demonstrate that standard molecular dynamics algorithms can run efficiently on both the CPU and an x86-based coprocessor using the same subroutines. As a consequence, we demonstrate that code optimizations for the coprocessor also result in speedups on the CPU; in extreme cases up to 4.7X. We provide results for LAMMAS benchmarks and for production molecular dynamics simulations using the Stampede hybrid supercomputer with both Intel (R) Xeon Phi (TM) coprocessors and NVIDIA GPUs: The optimizations presented have increased simulation rates by over 2X for organic molecules and over 7X for liquid crystals on Stampede. The optimizations are available as part of the "Intel package" supplied with LAMMPS. (C) 2015 Elsevier B.V. All rights reserved.
- Authors:
-
- Intel Corporation, Portland, OR (United States)
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Shell India Markets Private Limited, Bangalore (India)
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
- Publication Date:
- Research Org.:
- Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States). Oak Ridge Leadership Computing Facility (OLCF)
- Sponsoring Org.:
- USDOE Office of Science (SC)
- OSTI Identifier:
- 1261448
- Alternate Identifier(s):
- OSTI ID: 1246679
- Grant/Contract Number:
- AC05-00OR22725; AC04-94AL85000
- Resource Type:
- Accepted Manuscript
- Journal Name:
- Computer Physics Communications
- Additional Journal Information:
- Journal Volume: 195; Journal Issue: C; Journal ID: ISSN 0010-4655
- Publisher:
- Elsevier
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; Molecular dynamics; Xeon Phi; GPU; Coprocessor; Accelerator; Many-core; PERFORMANCE; POTENTIALS; MORPHOLOGY
Citation Formats
Michael Brown, W., Carrillo, Jan-Michael Y., Gavhane, Nitin, Thakkar, Foram M., and Plimpton, Steven J. Optimizing legacy molecular dynamics software with directive-based offload. United States: N. p., 2015.
Web. doi:10.1016/j.cpc.2015.05.004.
Michael Brown, W., Carrillo, Jan-Michael Y., Gavhane, Nitin, Thakkar, Foram M., & Plimpton, Steven J. Optimizing legacy molecular dynamics software with directive-based offload. United States. https://doi.org/10.1016/j.cpc.2015.05.004
Michael Brown, W., Carrillo, Jan-Michael Y., Gavhane, Nitin, Thakkar, Foram M., and Plimpton, Steven J. Thu .
"Optimizing legacy molecular dynamics software with directive-based offload". United States. https://doi.org/10.1016/j.cpc.2015.05.004. https://www.osti.gov/servlets/purl/1261448.
@article{osti_1261448,
title = {Optimizing legacy molecular dynamics software with directive-based offload},
author = {Michael Brown, W. and Carrillo, Jan-Michael Y. and Gavhane, Nitin and Thakkar, Foram M. and Plimpton, Steven J.},
abstractNote = {The directive-based programming models are one solution for exploiting many-core coprocessors to increase simulation rates in molecular dynamics. They offer the potential to reduce code complexity with offload models that can selectively target computations to run on the CPU, the coprocessor, or both. In our paper, we describe modifications to the LAMMPS molecular dynamics code to enable concurrent calculations on a CPU and coprocessor. We also demonstrate that standard molecular dynamics algorithms can run efficiently on both the CPU and an x86-based coprocessor using the same subroutines. As a consequence, we demonstrate that code optimizations for the coprocessor also result in speedups on the CPU; in extreme cases up to 4.7X. We provide results for LAMMAS benchmarks and for production molecular dynamics simulations using the Stampede hybrid supercomputer with both Intel (R) Xeon Phi (TM) coprocessors and NVIDIA GPUs: The optimizations presented have increased simulation rates by over 2X for organic molecules and over 7X for liquid crystals on Stampede. The optimizations are available as part of the "Intel package" supplied with LAMMPS. (C) 2015 Elsevier B.V. All rights reserved.},
doi = {10.1016/j.cpc.2015.05.004},
journal = {Computer Physics Communications},
number = C,
volume = 195,
place = {United States},
year = {Thu May 14 00:00:00 EDT 2015},
month = {Thu May 14 00:00:00 EDT 2015}
}
Web of Science
Works referenced in this record:
Implementing molecular dynamics on hybrid high performance computers – short range forces
journal, April 2011
- Brown, W. Michael; Wang, Peng; Plimpton, Steven J.
- Computer Physics Communications, Vol. 182, Issue 4
Implementing molecular dynamics on hybrid high performance computers – Particle–particle particle-mesh
journal, March 2012
- Brown, W. Michael; Kohlmeyer, Axel; Plimpton, Steven J.
- Computer Physics Communications, Vol. 183, Issue 3
An Evaluation of Molecular Dynamics Performance on the Hybrid Cray XK6 Supercomputer
journal, January 2012
- Michael Brown, W.; Nguyen, Trung D.; Fuentes-Cabrera, Miguel
- Procedia Computer Science, Vol. 9
Implementing molecular dynamics on hybrid high performance computers—Three-body potentials
journal, December 2013
- Brown, W. Michael; Yamada, Masako
- Computer Physics Communications, Vol. 184, Issue 12
Fast Parallel Algorithms for Short-Range Molecular Dynamics
journal, March 1995
- Plimpton, Steve
- Journal of Computational Physics, Vol. 117, Issue 1
A flexible algorithm for calculating pair interactions on SIMD architectures
journal, December 2013
- Páll, Szilárd; Hess, Berk
- Computer Physics Communications, Vol. 184, Issue 12
A generalized Gay-Berne intermolecular potential for biaxial particles
journal, April 1995
- Berardi, R.; Fava, C.; Zannoni, C.
- Chemical Physics Letters, Vol. 236, Issue 4-5
All-Atom Empirical Potential for Molecular Modeling and Dynamics Studies of Proteins †
journal, April 1998
- MacKerell, A. D.; Bashford, D.; Bellott, M.
- The Journal of Physical Chemistry B, Vol. 102, Issue 18
Quiet high-resolution computer models of a plasma
journal, February 1974
- Hockney, R. W.; Goel, S. P.; Eastwood, J. W.
- Journal of Computational Physics, Vol. 14, Issue 2
Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes
journal, March 1977
- Ryckaert, Jean-Paul; Ciccotti, Giovanni; Berendsen, Herman J. C.
- Journal of Computational Physics, Vol. 23, Issue 3
Molecular Dynamics Simulation of Dark-adapted Rhodopsin in an Explicit Membrane Bilayer: Coupling between Local Retinal and Larger Scale Conformational Change
journal, October 2003
- Crozier, Paul S.; Stevens, Mark J.; Forrest, Lucy R.
- Journal of Molecular Biology, Vol. 333, Issue 3
On the morphology of polymer-based photovoltaics
journal, March 2012
- Liu, Feng; Gu, Yu; Jung, Jae Woong
- Journal of Polymer Science Part B: Polymer Physics, Vol. 50, Issue 15
Development and testing of a general amber force field
journal, January 2004
- Wang, Junmei; Wolf, Romain M.; Caldwell, James W.
- Journal of Computational Chemistry, Vol. 25, Issue 9
Modelling of P3HT:PCBM interface using coarse-grained forcefield derived from accurate atomistic forcefield
journal, January 2014
- To, T. T.; Adams, S.
- Physical Chemistry Chemical Physics, Vol. 16, Issue 10
Liquid crystal nanodroplets in solution
journal, January 2009
- Brown, W. Michael; Petersen, Matt K.; Plimpton, Steven J.
- The Journal of Chemical Physics, Vol. 130, Issue 4
Rupture mechanism of liquid crystal thin films realized by large-scale molecular simulations
journal, January 2014
- Nguyen, Trung Dac; Carrillo, Jan-Michael Y.; Matheson, Michael A.
- Nanoscale, Vol. 6, Issue 6
Computational aspects of many-body potentials
journal, May 2012
- Plimpton, Steven J.; Thompson, Aidan P.
- MRS Bulletin, Vol. 37, Issue 5
Works referencing / citing this record:
Design considerations for GPU-aware collective communications in MPI: Design considerations for GPU-Aware collective communications in MPI
journal, May 2018
- Faraji, Iman; Afsahi, Ahmad
- Concurrency and Computation: Practice and Experience, Vol. 30, Issue 17
Accelerating AIREBO: Navigating the Journey from Legacy to High‐Performance Code
journal, February 2019
- Höhnerbach, Markus; Bientinesi, Paolo
- Journal of Computational Chemistry, Vol. 40, Issue 14
Scaling molecular dynamics beyond 100,000 processor cores for large‐scale biophysical simulations
journal, April 2019
- Jung, Jaewoon; Nishima, Wataru; Daniels, Marcus
- Journal of Computational Chemistry, Vol. 40, Issue 21
An Efficient Optimization of Hll Method for the Second Generation of Intel Xeon Phi Processor
journal, May 2018
- Kulikov, I. M.; Chernykh, I. G.; Glinskiy, B. M.
- Lobachevskii Journal of Mathematics, Vol. 39, Issue 4
Numerical Modeling of Hydrodynamic Turbulence with Self-gravity on Intel Xeon Phi KNL
book, August 2019
- Kulikov, Igor; Chernykh, Igor; Berendeev, Evgeny
- Parallel Computational Technologies: 13th International Conference, PCT 2019, Kaliningrad, Russia, April 2–4, 2019, Revised Selected Papers, p. 309-322
Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation
book, January 2019
- Sedova, Ada; Tillack, Andreas F.; Tharrington, Arnold
- Accelerator Programming Using Directives