Accelerating protein structure recovery using graphics processing units
ICCS'05 Proceedings of the 5th international conference on Computational Science - Volume Part I
Accelerated 2d image processing on GPUs
ICCS'05 Proceedings of the 5th international conference on Computational Science - Volume Part II
Implementing survey propagation on graphics processing units
SAT'06 Proceedings of the 9th international conference on Theory and Applications of Satisfiability Testing
Hi-index | 0.00 |
The primary purpose of this research is to accelerate the recovery of protein structures from x-ray crystallography data by using the GPU (graphics processing unit) of commonly available 3D graphics hardware as a vector processor performing matrix-based computations in distributed fashion with the CPU. In addition, this research outlines a reusable framework for scientific computing using the CPU and GPU together for vector computation. Finally, this research demonstrates a further benefit to scientists who employ the aforementioned algorithms and framework in the area of visualization. The motivating factors behind this research are that GPUs are powerful, underutilized, and inexpensive vector processors that are readily available and compatible with multiple scientific computing platforms. This research aims to make it feasible for scientists and researchers to upgrade their computers and commonly used software packages easily for very little cost and receive the maximum utilization benefit of their entire system's processing power. Beginning with a review of literature related to general-purpose computing on GPUs and literature on the specific computational tasks involved in bioinformatics research, a new hybrid GPU-CPU algorithm to accelerate the process of protein structure recovery from x-ray crystallography data is produced and applied. Finally, the results of this application are examined, a reproducible framework for additional research in applying this method is developed, and additional benefits of CPU-GPU utilization beyond the speed-up achieved in the computation-intensive steps are discussed.