Scalable molecular dynamics on CPU and GPU architectures withNAMD

• Published in: The Journal of chemical physics 2020-07, Vol.153 (4)
• Main Authors: Hardy, David J, Maia, Julio D, Stone, John E, Ribeiro, João V, Bernardi, Rafael C, Fiorin Giacomo, Hénin Jérôme, McGreevy, Ryan, Radak, Brian K, Skeel, Robert D, Singharoy Abhishek, Wang, Yi, Roux Benoît
• Format: Article
• Language: English
• Subjects: Algorithms; Charge distribution; Computer simulation; Electrostatics; External pressure; Molecular dynamics; Sampling; Simulation; Source code; Supercomputers
• ISSN: 0021-9606; 1089-7690
• DOI: 10.1063/5.0014475

NAMD is a molecular dynamics program designed for high-performance simulations of verylarge biological objects on CPU- and GPU-based architectures. NAMD offers scalableperformance on petascale parallel supercomputers consisting of hundreds of thousands ofcores, as well as on inexpensive commodity clusters commonly found in academicenvironments. It is written in C++ and leans on Charm++ parallel objects for optimalperformance on low-latency architectures. NAMD is a versatile, multipurpose code thatgathers state-of-the-art algorithms to carry out simulations in apt thermodynamicensembles, using the widely popular CHARMM, AMBER, OPLS, and GROMOS biomolecular forcefields. Here, we review the main features of NAMD that allow both equilibrium andenhanced-sampling molecular dynamics simulations with numerical efficiency. We describethe underlying concepts utilized by NAMD and their implementation, most notably forhandling long-range electrostatics; controlling the temperature, pressure, and pH;applying external potentials on tailored grids; leveraging massively parallel resources inmultiple-copy simulations; and hybrid quantum-mechanical/molecular-mechanicaldescriptions. We detail the variety of options offered by NAMD for enhanced-samplingsimulations aimed at determining free-energy differences of either alchemical orgeometrical transformations and outline their applicability to specific problems. Last, wediscuss the roadmap for the development of NAMD and our current efforts toward achievingoptimal performance on GPU-based architectures, for pushing back the limitations that haveprevented biologically realistic billion-atom objects to be fruitfully simulated, and formaking large-scale simulations less expensive and easier to set up, run, and analyze. NAMDis distributed free of charge with its source code at www.ks.uiuc.edu.