一种基于GPU的实时水波模拟方法

大范围动态水面场景的实时三维仿真是虚拟海洋环境以及三维游戏中研究的难点和热点。由于水的动态性以及各种复杂的光照等效果,传统基于CPU的模拟很难达到实时交互的能力。近年来图形处理器(GPU)性能得到大幅度提高,尤其引人注意的是出现可编程特性,这使得人们可以充分利用GPU上提供的运算器来设计算法完成一定的任务。基于此,针对目前实时水波模拟的瓶颈及提倡的"解放CPU"(Offload CPU)的思想,本文对水波模拟的模型进行了研究,提出了一种基于GPU的水波模拟方法,充分利用GPU的并行运算及浮点运算能力.加快水面模拟的速度。文章给出了一定网格大小下,数量不同的点波源模拟水面的效果,并给出了单纯利用CPU和基于GPU两种方法的对比情况。

A method of simulation of real-time water based on GPU

The real-time 3D simulation of large range and dynamic water has become the focus and difficulty to which the visual ocean environment and 3D game are now paying more attention.Due to dynamic of water and many complicated effects of illumination, and so on,it's difficult to achieve good frame rate in real-time alternation.The performance of Graphics Processing Unit(GPU) has improved greatly recently,what's remarkable is the programmable characteristic,that can be used to complete some work by desig- ning arithmetics in GPU.So,aimed at the bottleneck of simulation of real-time water and"Offload CPU"proposed at present,this pa- per studies on the model of water,and brings toward a simulation method of water based on GPU.which makes the best of the GPU' s ability of parallel and floating-point operation,and can actually improve the frame rate on simulation of water.It presents the effect under certain sizing grid and different-number spot wave source,and gives the contrast situation between using CPU purely and based on the GPU.