Abstract: In order to improve the computational efficiency of high-frequency acoustic scattering in underwater targets, a Kirchhoff approximate integral calculation model based on graphics processing unit (GPU) parallel computation is established in this paper. Firstly, a parallel model based on GPU thread allocation is established for the constant element model and the surface element exact integral model of the Kirchhoff approximate integral method for target sound scattering, forming a parallel algorithm model. Then, taking a rigid sphere with a radius of 1m as the target, the GPU parallel model is used to calculate the target intensity of sound scattering, and the accuracy of the algorithm is verified by comparing it with the analytical solution. Finally, using the Benchmark model as the target, the acceleration ratio of the GPU parallel computing model is compared and analyzed by simulating the intensity of acoustic scattering under different conditions. The results show that the GPU parallel computing efficiency of the constant element model is 4-5 times higher than that of traditional serial computing. The GPU parallel computing efficiency of the planar element precise integration model is 8-11 times higher than that of traditional serial computing. The parallelization mode based on GPU has a significant acceleration effect on the computational efficiency of the Kirchhoff approximate integral method for target sound scattering, and the advantage of GPU computation becomes more obvious as the number of surface elements increases.