利用CUDA快速计算三角网格模型简化误差

摘要

Propose a method to evaluate the overall simplification error of triangular meshes. It considers the contribution of all sample points to the error, which is more precise than the former methods that only consider the maximum error. Achieve a highly paralleled algorithrn to calculate the simplification error of triangular meshes by using CUDA. Uniform gird is adopted to manage triangles of each mesh to accelerate the space search speed. Proper data structure is designed to solve the problems such as CUDA does not support pointer, dynamically memory allocating and to avoid synchronization as much as possible. Finally, experimental results show the efficiency and effectiveness of this algorithm, which verify the advantages of GPU on high performance calculation compared to CPU.%提出从整体上考虑三角网格模TJ的简化误差的方法.该方法综合考虑了三角网格模型上所有采样点对误差的贡献,较以往方法只考虑最大误差更加准确.利用CUDA实现了高度并行化的网格模型简化误差计算算法.算法采用平均单元格来组织三角片,以便快速空间查询.同时设计了相应的数据结构,克服了CUDA没有指针、不能动态申请资源、尽量避免同步操作等问题.最后通过实验证明了本算法在速度和数据量上的优越性,并且阐述了GPU相对于CPU在高性能计算上的优点.