Explicit design of FPGA-based coprocessors for short-range force computations in molecular dynamics simulations

Abstract

FPGA-based acceleration of molecular dynamics simulations (MD) has been the subject of several recent studies. The short-range force computation, which dominates the execution time, is the primary focus. Here we combine: a high level of FPGA-specific design including cell lists, systematically determined interpolation and precision, handling of exclusion, and support for MD simulations of up to 256K particles. The target system consists of a standard PC with a 2004-era COTS FPGA board. There are several innovations: new microarchitectures for several major components, including the cell list processor and the off-chip memory controller; and a novel arithmetic mode. Extensive experimentation was required to optimize precision, interpolation order, interpolation mode, table sizes, and simulation quality. We obtain a substantial speed-up over a highly tuned production MD code.

Efficient particle-pair filtering for acceleration of molecular dynamics simulation

“…In our previous work we completely overlap the shortrange force computation with host work; in [2] we found host overhead to be around 40ms on a single core. This code must now be parallelized to keep it off of the critical path.…”

“…The long-range component of the Coulombic force is the long-range force. Since the long-range force computation is generally a small fraction of the total [2,4], our focus here is on the short-range force. Given a particle pair i and j:…”

“…In particular, we study the filtering of particle pairs with nearly zero mutual force, a standard optimization in MD codes. When coupled with our recent work on the optimization of MD force pipelines [1], this has the potential to improve performance of our previous full MD implementation [2] four-fold. The issue of filtering particle pairs emerges from the geometric mismatch between two shapes: (i) the cubes (or other polyhedrons) into which it is convenient to partition the simulation space and (ii) the spheres around each particle in which the short-range force is non-zero.…”

Molecular Dynamics Simulation of Iron — A Review

“…167 A notable method to solve this problem is the Particle mesh Ewald (PME), 168 which relies on computing the force in the reciprocal domain using the three-dimensional fast Fourier transform (FFT), yet the implementation in FPGA is less ine±cient. 169 The multi-grid approach is also a possible approach, whose sequential processing can attain roughly the same computational speed as the PME. 170 A number of attempts have been made to implement the computation of the Coulombic force, see for example Ref.…”

Implementation of a field programmable gate array for wireless control of a lab-on-a-robot

“…Currently, FPGA modules have progressed in performance and customizability for system connectivity, and enable having logic executing and a microprocessor running simultaneously. Different applications [17][18][19][20][21], such as digital signal processing, computer hardware emulation, robotics, and embedded control systems have benefited from the implementation of FPGA boards. Additionally, FPGAs offer more than one dedicated universal asynchronous receiver/transmitter (UART) and wide-ranging hardware and software interface support, such as Peripheral Component Interface (PCI) and Universal Serial Bus (USB).…”