Abstract: A new one-dimensional Godunov-type hydrodynamic model based on GPU heterogeneous parallel computing was developed to overcome two key limitations of existing models:the absence of a unified discharge boundary treatment for supercritical and subcritical flows,and the need for higher computational efficiency.The model introduces two main innovations:(1)Numerically,a novel discharge boundary treatment method based on a “source-term” approach is proposed,which strictly ensures global conservation of mass and momentum and provides a unified framework for handling discharge boundaries under supercritical and subcritical conditions;(2)Computationally,the model is implemented using GPU heterogeneous parallel computing technology with the CUDA architecture,enabling high-density parallel acceleration.The accuracy and efficiency of the model were verified using three classic test cases.Results show that the numerical solutions obtained with the new model agree closely with analytical solutions，and an order-of-magnitude speedup is achieved by using the GPU heterogeneous parallel computing technology,with the acceleration effect becoming more pronounced as the number of computational units increases,confirming the successful development of the model.Finally,the model was successfully applied to the Chengjiang River reach in Guangxi.