DEM-1: A Particle Simulation Machine for Efficient Short-Range Interaction Computations
نویسندگان
چکیده
This paper describes the architecture of high performance, particle simulation machine, DEM-1 for short-range particle interaction computations. All existing particle simulation machines have specialized pipelines to calculate long-range particle interactions effectively. However, their ability to perform particle simulations efficiently diminishes with short-range interactions. The communication cost component of particle simulations will play a significant role in performance when the computation cost becomes O(N). DEM-1's three dimensional torus high-speed network reduces the communication cost while 2048 local processors perform the time integration. The elimination of pipeline bubbles in DEM-1 is achieved by specially designed cut off judgment units. Each specialized pipeline consists of dedicated data path supported by position vector prefetch dual ported memory. The performance of DEM-1 will be presented with very large-scale Embedded Atom Method (EAM) molecular dynamics simulations.
منابع مشابه
Error Analysis of the Poisson PM Force Field Scheme for Particle-Based Simulations of Biological Systems
The particle-particle-particle-mesh (PM) force field scheme [1], [2] is an efficient approach to account for both the long-range and short-range electrostatic interactions in the simulation of biological systems. Resolving the particle-mesh term in real-space [2], [3] improves the original approach of Hockney et al. [1] by allowing for the inclusion of non-periodic boundary conditions. To date,...
متن کاملGPU parallel simulation algorithm of Brownian particles with excluded volume using Delaunay triangulations
A novel parallel simulation algorithm on the GPU, implemented in CUDA and C++, is presented for the simulation of Brownian particles that display excluded volume repulsion and interact with long and short range forces. When an explicit Euler-Maruyama integration step is performed to take into account the pairwise forces and Brownian motion, particle overlaps can appear. The excluded volume prop...
متن کاملQuantitative Analysis of Accuracy of Voidage Computations in CFD-DEM Simulations
CFD-DEM (Computational Fluid Dynamics – Discrete Element Modelling) is a twophase flow numerical modelling technique, where the Eulerian method is used for the fluid and the Lagrangian method for the particles. The two phases are coupled by a fluid-particle interaction force (i.e. drag force) which is computed using a correlation. In a two-phase flow, one critical parameter is the voidage (or v...
متن کاملBack-calculation of mechanical parameters of shell and balls materials from discrete element method simulations
Discrete Element Method (DEM) is extensively used for mathematical modeling and simulating the behavior of discrete discs and discrete spheres in two and three dimensional space, respectively. Prediction of particles flow regime, power draw and kinetic energy for a laboratory or an industrial mill is possible by DEM simulation. In this article, a new approach was used to assess the main paramet...
متن کاملOn the Efficient Simulation of Electron-electron Interactions in Nanoscale Mosfets
The ongoing scaling ofMOSFET devices requires newmodels and simulation techniques for particle based Monte Carlo simulations of devices in the nanoscale regime. Electron-electron interactions must be considered because of two reasons. First, the Coulomb force is a long range force implying that both the short and the long range interactions have to be considered. Second, the hot-carrier and sho...
متن کامل