Molecular Dynamics on FPGA Based Accelerated Processing Units

Abstract

Abstract. One of the main problems in providing the amount of computation requested by the Molecular Dynamic domain is to offer an appropriate architectural environment for solving all the aspects of the intense parts of the involved computation. Current solutions accelerate only partially the intense computation -forces computation & position and speed updates, which represents around 75% from the total computational effort -thus limiting the help provided by the parallel computing resources involved. The aim of this paper is to introduce a parallel accelerator featured with functions able to add to the accelerated functions the neighbourhood list building, which represent around 25% from the total computation. Thus, accelerations higher than the current ∼ 4× are expected. Our proposal, the MapReduce Accelerator, is evaluated using the Gromacs system. The Martini water example, running on a cycle accurate simulator, is used to evaluate the speed-up and the energy.

On the Feasibility of FPGA Acceleration of Molecular Dynamics Simulations

“…Most studies on FPGA accelerated MD target the PP calculations [10], [11], [13], [14], [20][21][22], since they make up for over 90% of the sequential runtime in typical simulations. Apart from these studies, only relatively few papers cover PME and other aspects [16], [18], [19], [45].…”

“…Besides these MPI and GPU-based solutions, FPGA accelerators have repeatedly been proposed as a viable alternative to accelerate the compute intensive parts [10][11][12][13][14][15][16][17][18][19][20][21][22] . However, these studies only show estimated or measured speedup with respect to older CPU implementations.…”