Tinker-HP : Accelerating Molecular Dynamics Simulations of Large Complex Systems with Advanced Point Dipole Polarizable Force Fields using GPUs and Multi-GPUs systems Olivier Adjoua,y Louis Lagardère*,y,z Luc-Henri Jolly,z Arnaud Durocher,{ Thibaut Very,x Isabelle Dupays,x Zhi Wang,k Théo Jaffrelot Inizan,y Frédéric Célerse,y,? Pengyu Ren,# Jay W. Ponder,k and Jean-Philip Piquemal∗,y,# ySorbonne Université, LCT, UMR 7616 CNRS, F-75005, Paris, France zSorbonne Université, IP2CT, FR2622 CNRS, F-75005, Paris, France {Eolen, 37-39 Rue Boissière, 75116 Paris, France xIDRIS, CNRS, Orsay, France kDepartment of Chemistry, Washington University in Saint Louis, USA ?Sorbonne Université, CNRS, IPCM, Paris, France. #Department of Biomedical Engineering, The University of Texas at Austin, USA E-mail:
[email protected],
[email protected] arXiv:2011.01207v4 [physics.comp-ph] 3 Mar 2021 Abstract We present the extension of the Tinker-HP package (Lagardère et al., Chem. Sci., 2018,9, 956-972) to the use of Graphics Processing Unit (GPU) cards to accelerate molecular dynamics simulations using polarizable many-body force fields. The new high-performance module allows for an efficient use of single- and multi-GPUs archi- 1 tectures ranging from research laboratories to modern supercomputer centers. After de- tailing an analysis of our general scalable strategy that relies on OpenACC and CUDA , we discuss the various capabilities of the package. Among them, the multi-precision possibilities of the code are discussed. If an efficient double precision implementation is provided to preserve the possibility of fast reference computations, we show that a lower precision arithmetic is preferred providing a similar accuracy for molecular dynamics while exhibiting superior performances.