研究了图形处理器(GPU)计算液晶自适应波前重构的并行性。介绍了液晶自适应光学的Zernike模式波前重构算法,论述了GPU的通用架构和GPU实现波前重构的方法。在此基础上提出了利用GPU拥有的RGBA4个颜色通道进行并行计算,进一步加快计算速度,最后给出了实验结果。结果表明:在GPU计算波前重构时,利用RGBA颜色通道的并行计算,将计算速度提高了3倍多。
A wavefront reconstruction parallel computing based on the GPU is introduced in this paper. A Zernike modal wavefront reconstruction algrithm based on liquid crystal adaptive optics is introduced. The general model of GPU and the wavefront reconstruction method using GPU is discussed; based on it, the parallel calculation with four color channels of GPU will increase the calculational speed. The effect and speed about wavefront reconstruction method based on GPU is tested and compared. As a result, the parallelism enjoyed by the GPU processing works largely and provides a computational power 3 times greater than the GPU direct computing.