GPU通用计算模式在岩土工程中的应用

由于岩土工程地质条件的复杂性及其规模的不断增大，对大规模数值计算速度的要求越来越高。显卡核心单元（GPU）由于其硬件构造特殊，有着并行计算上的独特优势、高速浮点运算性能和超高的内存带宽，可以很好地解决大规模的科学计算速度问题。文中介绍了GPU与CPU的硬件构架差异，总结了多核CPU、工作站等方式发展的局限性及GPU在并行运算方面的优势，详细阐述了GPU各类计算模式的发展特点及其成果，展示了其在坝区渗透特性中随机微分方程加速求解过程中的优越性，探讨了采用GPU进行大规模岩土工程数值计算的应用前景。

Application of general-purpose computation on GPUs to geotechnical engineering

Because of the complex geological conditions of geotechnical engineering and its increasing project scale, the requirements of the calculation speed for large-scale numerical simulation become more strict. Graphics processing unit (GPU), the core unit of graphics card, can solve this problem with the advantages of parallel computing, high-speed floating-point performance and high memory bandwidth. The hardware architecture differences between GPU and CPU are analyzed firstly. Then the advantages and disadvantages of CPU and GPU are summarized; the development and achievement of general-purpose computation on GPU are described in detail. Finally, an example about application of GPU to stochastic differential equations of permeability is given; the advantage of GPU is displayed and the application of GPU computing to geotechnical engineering is discussed.