论文部分内容阅读
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。
Due to the complexity of geotechnical geological conditions and its increasing size, the requirements for large-scale numerical calculation speed are getting higher and higher. Due to its special hardware structure, GPU core unit (GPU) has the unique advantages of parallel computing, high-speed floating-point arithmetic performance and high memory bandwidth, which can well solve the large-scale scientific computing speed problem. This paper introduces the hardware architecture differences between GPU and CPU, summarizes the limitations of multi-core CPUs and workstations, and the advantages of GPU in parallel computing. It also expounds the development characteristics and achievements of various GPU computing modes in detail, In the dam zone permeability characteristics of stochastic differential equations in the process of accelerating the superiority of the use of GPU for large-scale numerical calculation of geotechnical engineering prospects.