A molecular docking system using CUDA
Konkuk University, Seoul, Korea
In ICHIT ’09: Proceedings of the 2009 International Conference on Hybrid Information Technology (2009), pp. 28-33.
@conference{roh2009molecular,
title={A molecular docking system using CUDA},
author={Roh, Y. and Lee, J. and Park, S. and Kim, J.I.},
booktitle={Proceedings of the 2009 International Conference on Hybrid Information Technology},
pages={28–33},
year={2009},
organization={ACM}
}
A Molecular Docking System enables biologists to check whether two molecular models can be combined at a specific position and remain in their stable states by simulation. It can be used in developing new materials and designing new drugs. Since the docking simulation consists of several complicated computations at the level of atoms, it requires high computing capabilities such as super computers and parallel computing systems. We propose a molecular docking system using parallel GPUs in this paper. In our proposed method, a GPU can process an equation as a single logical work unit. The computations can be executed through parallel GPUs in real-time. The proposed system was evaluated in its performance by comparing with conventional CPU-based systems. A series of experiments for measuring performance of the system showed that our system is 33 to 287 times faster than the CPU-based systems.
November 9, 2010 by hgpu
