In ground-based solar observation, when the light passes through atmosphere, it will be affected by atmospheric turbulence which will cause translation, distortion and blurring of the received image. In order to eliminate or reduce the effects of atmospheric turbulence, post-image processing technology was used to obtain high-resolution solar images. High-resolution image can be obtained by post-facto reconstruction algorithm based on the speckle interferometry and speckle masking. However, due to its complex calculation, the algorithm can not meet the requirement of real-time detection. On the basis of the principle of the algorithm, solar speckle image reconstruction algorithm has been parallelized by using CUDA parallel computing architecture in this research. The experimental results show that a 2304 pixel×1984 pixel image of TiO channel can be reconstructed within 70 s under GPU environment. Compared with the program running on CPU, the speed-up radio can up to 7.