Multiple Prediction through Inversion - GPU Acceleration

Multiple prediction through inversion (MPI) is an effective method for the free-surface multiple attenuation. Due to the large amount of matrix operations involved, the computational cost of MPI scheme is considerably expensive. In this paper, we apply the technology of Graphics Processor Unit (GPU) into the implementation of MPI scheme. As GPU has highly parallel structure and is suitable for matrix related calculations, we can improve the computational efficiency dramatically when applying GPU for the free-surface multiple prediction with MPI scheme.