Computer simulation of liquids
Computer simulation of liquids
Understanding Molecular Simulation
Understanding Molecular Simulation
Using OpenMP: Portable Shared Memory Parallel Programming (Scientific and Engineering Computation)
Using OpenMP: Portable Shared Memory Parallel Programming (Scientific and Engineering Computation)
The Art of Multiprocessor Programming
The Art of Multiprocessor Programming
Parallelizing molecular dynamics solutions for high performance
Proceedings of the ATIP/A*CRC Workshop on Accelerator Technologies for High-Performance Computing: Does Asia Lead the Way?
| Hi-index | 0.00 |
Parallel and distributed computing techniques have been applied to solve problems that have long running times. Molecular dynamics (MD) is used extensively in physics, chemistry, material sciences and biology to study statistical mechanical properties of systems. MD simulators are known to have extremely long running times, especially as the number of particles being simulated increase. Researchers have used PVM and MPI techniques to speed up MD computations and have achieved 5X to 10X speed up using clusters of computers. Recently NVIDIA and other Graphics Processing Unit (GPU) manufacturers have made available GPU cards for general purpose computing besides rendering graphics. We present a novel idea of solving the MD simulation problem in parallel using General Purpose Graphics Processing Unit (GPGPU) cards. MD problems are particularly amenable to data parallelization. Exploiting this property of the problem and the parallel constructs of GPGPU libraries, we present a parallel MD algorithm, which has a peak speedup of 200X for large problem sizes. Our work can be generalized to parallelize the now popular MapReduce programming paradigm.