swMD: Performance Optimizations for Molecular Dynamics Simulation on Sunway Taihulight

Molecular dynamics is an extensively utilized computational tool for solids, liquids and molecules simulation. Currently, much research on molecular dynamics simulation focuses on simplifying forces or parallelizing tasks to reduce the overheads of forces computation. However, the molecular dynamics simulation still remains challenging since the communication and neighbor list construction are time-consuming in the existing algorithm. In this paper, we propose a swMD optimization strategy including a new communication mode called ghost communication to reduce superfluous communication overheads and an innovative neighbor list algorithm to improve the construction efficiency of it. Moreover, we accelerate computation by utilizing many-core resources on Sunway Taihulight and present an auto-tuning Producer-Consumer pairing algorithm to make neighbor list construction happen in fast register communication. Compared to traditional methods, swMD optimization strategy obtains a maximal 82.2% and an average of 79.4% performance improvement. We also evaluate the scalability up to 266,240 cores and the results demonstrate the high efficiency of swMD optimization strategy on communication, computation and neighbor list construction respectively.

[1]  Kun Li,et al.  FastNBL: fast neighbor lists establishment for molecular dynamics simulation based on bitwise operations , 2019, The Journal of Supercomputing.

[2]  Carsten Kutzner,et al.  GROMACS 4:  Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. , 2008, Journal of chemical theory and computation.

[3]  Michela Taufer,et al.  FENZI: GPU-Enabled Molecular Dynamics Simulations of Large Membrane Regions Based on the CHARMM Force Field and PME , 2011, 2011 IEEE International Symposium on Parallel and Distributed Processing Workshops and Phd Forum.

[4]  J. Tersoff,et al.  Empirical interatomic potential for silicon with improved elastic properties. , 1988, Physical review. B, Condensed matter.

[5]  Joshua A. Anderson,et al.  General purpose molecular dynamics simulations fully implemented on graphics processing units , 2008, J. Comput. Phys..

[6]  Klaus Schulten,et al.  High-performance scalable molecular dynamics simulations of a polarizable force field based on classical Drude oscillators in NAMD. , 2011, The journal of physical chemistry letters.

[7]  Steve Plimpton,et al.  Fast parallel algorithms for short-range molecular dynamics , 1993 .

[8]  Kenli Li,et al.  Performance-Aware Model for Sparse Matrix-Matrix Multiplication on the Sunway TaihuLight Supercomputer , 2019, IEEE Transactions on Parallel and Distributed Systems.

[9]  William A. Goddard,et al.  Silicon nanowires as efficient thermoelectric materials , 2008, Nature.

[10]  Felix Höfling,et al.  Highly accelerated simulations of glassy dynamics using GPUs: Caveats on limited floating-point precision , 2009, Comput. Phys. Commun..

[11]  Sharon C. Glotzer,et al.  The development and expansion of HOOMD-blue through six years of GPU proliferation , 2013, 1308.5587.

[12]  Changjun Hu,et al.  Massively Scaling the Metal Microscopic Damage Simulation on Sunway TaihuLight Supercomputer , 2018, ICPP.

[13]  Weifeng Liu,et al.  swSpTRSV: a fast sparse triangular solve with sparse level tile layout on sunway architectures , 2018, PPoPP.

[14]  Leonard M. Blumenthal,et al.  Theory and applications of distance geometry , 1954 .

[15]  Gui-Rong Liu,et al.  Improved neighbor list algorithm in molecular simulations using cell decomposition and data sorting method , 2004, Comput. Phys. Commun..

[16]  L. Verlet.  Computer "Experiments" on Classical Fluids. I. Thermodynamical Properties of Lennard-Jones Molecules , 1967 .

[17]  William D. Mattson,et al.  Near-neighbor calculations using a modified cell-linked list method , 1999 .

[18]  William A. Goddard,et al.  Silicon Nanowires as Efficient Thermoelectric Materials. , 2008 .

[19]  Pak Lui,et al.  Strong scaling of general-purpose molecular dynamics simulations on GPUs , 2014, Comput. Phys. Commun..

[20]  Morten Hjorth Jensen,et al.  COMPUTATIONAL Physics , 2015 .

[21]  Klaus Schulten,et al.  Accelerating Molecular Modeling Applications with GPU Computing , 2009 .

[22]  Tao Li,et al.  CASpMV: A Customized and Accelerative SpMV Framework for the Sunway TaihuLight , 2021, IEEE Transactions on Parallel and Distributed Systems.

[23]  Sharon C. Glotzer,et al.  Efficient neighbor list calculation for molecular simulation of colloidal systems using graphics processing units , 2016, Comput. Phys. Commun..

[24]  George E. Karniadakis,et al.  Accelerating dissipative particle dynamics simulations on GPUs: Algorithms, numerics and applications , 2013, Comput. Phys. Commun..

[25]  Peng Zhang,et al.  Towards Highly Efficient DGEMM on the Emerging SW26010 Many-Core Processor , 2017, 2017 46th International Conference on Parallel Processing (ICPP).