基于CUDA的Bezier曲线生成算法并行化研究

Research on the Parallelization of Bezier Curve Generation Algorithm Based on CUDA

摘要: 针对Bezier曲线在几何造型系统和图形软件中的重要作用，提出了一种基于CUDA的对Bezier曲线进行计算的方法。在分析CUDA异构计算架构和Bezier曲线采样点计算独立性的基础上，在GPU上开启多线程对Bezier曲线进行计算。并根据控制顶点在采样点计算过程中不改变的特性，在存储器层次对采样点的计算进行优化。实验结果表明，基于CUDA改进后的Bezier曲线生成算法性能有了显著的提高。

Abstract: The Bezier curve is essential for geometric modeling system and graphic software. A method of calculating Bezier curve based on CUDA is proposed. Based on the analysis of the CUDA isomeric computing architecture and the independence of the Bezier curve sampling points, the multiple threads were opened on the GPU to calculate the Bezier curve. According to the characteristic that the control vertex does not change during the calculation of the sampling point, the calculation of the sampling point was optimized at the memory level. The experimental results show that the performance of Bezier curve generation algorithm based on CUDA has been improved remarkably.