RUS  ENG
Full version
JOURNALS // Numerical methods and programming // Archive

Num. Meth. Prog., 2009 Volume 10, Issue 2, Pages 158–175 (Mi vmp367)

This article is cited in 1 paper

Вычислительные методы и приложения

Parallel molecular dynamics with Ewald summation and integration on GPU

A. S. Boyarchenkova, S. I. Potashnikovb

a Institute of Mathematics of the Ural Branch of RAS
b Ural State Technical University, Ekaterinburg

Abstract: The problem of molecular dynamics simulation with periodic boundary conditions is considered. Efficient algorithms for the Ewald summation method are discussed. Parallel implementations on GPU with NVIDIA CUDA technology are proposed. The integration of motion equations on GPU with single and double precision of floating point arithmetic is analyzed. On an NVIDIA GeForce GTX 280 video card, a remarkable speed-up up to 890 times is achieved compared to a scalar code running on Intel Core 2 Quad Q9550 CPU.

Keywords: molecular dynamics; Ewald summation; parallel computing; graphics processors; CUDA.

UDC: 004.021:004.94:519.683:519.684:544.223



© Steklov Math. Inst. of RAS, 2024