Abstract: Water conservancy calculation is the basis of water conservancy planning, design and operation. Improving the efficiency of water conservancy calculation is of great practical significance for water informatization and water resources management. The technical comparison shows that GPU（Graphics Processing Unit） parallel optimization technology is a cost-effective speed-up strategy. This paper systematically summarizes the application progress of GPU parallel optimization in water conservancy calculation, briefly introduces several parallel techniques currently used in the proposed calculation; constructively puts forward the application prospects and advantages of this technology in reservoir scheduling, mid-and?long term hydrologic forecasting and hydrological model calculation; summarizes the general methods with the application of this technology and provides guidance for technology promotion. Finally, from the perspective of discipline development and application demand, the difficulties and future trends of technology application are pointed out, in order to provide reference for the application of GPU parallel optimization technology in water conservancy calculation.

Key words: water conservancy calculation, Graphics Processing Unit（GPU）, parallel computing, hydrodynamic simulation, wisdom water, depth learning, big data

摘要： 水利计算是水利规划、设计以及运行的基础，提高水利计算的效率对水利信息化和水资源管理具有重要实践意义。通过技术对比发现，GPU（Graphics Processing Unit）并行优化技术是性价比较高的提速策略。系统概述了GPU并行优化技术在水利计算中的应用进展；简要介绍了当前应用较多的几种并行技术；建设性提出了该项技术在水库调度、中长期水文预报和水文模型计算中的应用前景和优势；详细总结了应用该项技术的一般方法，为技术推广提供指导。最后从学科发展和应用需求的角度，有针对性的提出了技术应用难点和今后发展趋势，以期为GPU并行优化技术在水利计算中的应用提供借鉴。

关键词: 水利计算, GPU, 并行计算, 水动力模拟, 智慧水务, 深度学习, 大数据