| 基于GPU的隐式不可压缩SPH流体模拟 |
| |
| 作者单位： | ;1.河南广播电视大学继续教育学院;2.郑州航空工业管理学院;3.河南广播电视大学信息工程系;4.电子科技大学计算机科学与工程学院 |
| |
| 摘 要： | 提出一种基于图形处理单元(Graphic Processing Unit,GPU)的不可压缩流体并行模拟算法.该算法使用并行基数排序技术提升了邻居查找效率,同时使用了GPU上的片上高速共享存储器,将流体计算过程中所需用到的数据尽可能从GPU的全局存储器中拷贝至共享存储器中,减小数据访问延迟,提高模拟效率.实验结果表明,基于GPU的并行模拟算法可以大幅提高流体模拟程序的性能,与基于CPU的单线程实现相比,可以到达38.2倍的加速比.
|
| 关 键 词： | 基于物理的流体模拟 光滑粒子流体动力学 图形处理单元 不可压缩性 |
GPU-basedImplicit Incompressible SPH Fluid Simulation |
| |
| Affiliation: | ,The School of Continuing Education,Henan Radio & Television University,Zhengzhou Institute of Aeronautical Industry Management,The Department of Information Engineering,Henan Radio & Television University,School of Computer Science and Engineering,University of Electronic Science and Technology of China |
| |
| Abstract: | We present a Graphic Processing Unit(GPU)Based parallel simulation algorithm for incompressible fluids.The algorithm utilizes parallel radix sort technique to improve the efficiency of neighbor search and takes full advantage of GPU's fast on-chip shared memory.The fluid data is copied from GPU's global memory to shared memory,which reduces the latency of data access and thus improves simulation performance.The experimental result shows that the GPU-based parallel fluid simulation algorithm can largelyimprove the physics simulation performance.Compared to single-threaded CPU implementation,the GPU-based counterpart can achieve a 38.2speedup. |
| |
| Keywords: | physically-based fluid simulation Smoothed Particle Hydrodynamics Graphic Processing Unit incompressibility |
| 本文献已被 CNKI 等数据库收录！ |
|
