基于OpenCL加速的SPH流体仿真  

Fluid Simulation of SPH Based on Open CL

作　　者：肖苗苗[1] 刘箴[1] 史佳宾 刘婷婷[1] 刘翠娟[1] 刘邦权[1] 

机构地区：[1]宁波大学信息科学与工程学院,宁波315211

出　　处：《系统仿真学报》2015年第4期803-809,共7页Journal of System Simulation

基　　金：国家自然科学基金(61373068);浙江省自然科学基金(LY13F020037);宁波市科技计划项目(2013D10011;2014C50018);高等学校博士学科点专项科研基金(No.20133305110004);浙江省教育厅科研项目(Y201431792)

摘　　要：流体在虚拟现实场景中扮演重要角色,基于物理模型的流体动画可以增强数值仿真的真实感。但通常的基于物理模型的流体动画渲染,如果没有采用并行计算,不能达到实时模拟的效果。使用异构计算平台Open CL对SPH(smoothed particle hydrodynamics)方法进行加速,充分利用了Open CL与图形渲染API的数据互通性,并采用了CPU和GPU的协同工作的技术,能够发挥硬件潜力;同时,采用库伦定律和SPH方法计算表面张力和粒子法线,初步的实验结果表明该方法有希望用于大规模流体仿真。Fluid plays an important role in the Virtual Reality Scene. Physics-based fluid animation can enhance the reality of numerical simulation. The efficiency of fluid animation based on physical computing usually is not real-time rendering without parallel computing. Fluid simulation was proposed based on SPH using Open CL, which could make full use of the data interoperability of Open CL and the graphics API, and use the coordination work of CPU and GPU to increase the potential of the hardware; when physics computing, the surface particles and normals based on the Coulomb's law were solved out. The preliminary experiment shows that the method is hopeful in simulation a large scale fluid simulation.

关 键 词：流体仿真 并行计算 OPEN CL SPH 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]