ADVANCES IN GRAPHICS PROCESSING UNITS’APPLICATIONS TO THE COMPUTATION OF LARGE-SCALE MECHANICAL PROBLEMS
-
摘要: 现代图形处理器~(graphics processing units,GPU)具有较强的并行数值运算功能. 该文简单介绍了GPU的硬件结构,基于GPU通用计算的数据结构和实现方法,以及用于编写片元程序的OpenGL着色语言.介绍了应用GPU计算大规模力学问题的研究进展. 简要介绍了以下内容:应用GPU模拟自然界的流体现象,其实质是使用有限差分法求解Navier-Stokes方程;应用GPU实现有限元法计算, 使用基于GPU的共轭梯度法求解有限元方程组;应用GPU实现分子动力学计算, 用GPU计算原子间短程作用力,并生成邻近原子列表; 应用GPU实现量子力学Monte Carlo计算;应用GPU实现$n$个物体的引力相互作用,用GPU纹理存储$n$个物体的位置、质量、速度和加速度等.对基于图象处理器和中央处理器的计算作比较,已完成了以下基于GPU的计算:实现求解线性方程组的高斯消元法和共轭梯度法,并应用于大规模的有限元计算; 加速无网格法计算;加速线性和非线性分子结构力学方法计算;用于计算分析碳纳米管的力学性能.指出GPU在大规模力学计算中的研究方向.Abstract: In this paper, we simply describe the hardware architectures and properties of modern graphics processing units (GPUs). The GPUs’ applications on mechanics are presented. We have described the fundamental theories and methods on how to accelerate the computations on such mechanical problems as fluid simulation, finite element method, molecular dynamics, quantum Monte Carlo model, and gravitational N－body simulation with GPUs as co-processors. The limitations of general purpose computation on GPUs are analyzed. At last, we have presented the future work on mechanics using GPUs as co-processors.
-
计量
- 文章访问数: 2480
- HTML全文浏览量: 132
- PDF下载量: 1546
- 被引次数: 0
下载: