基于CUDA的地震倾角方位角并行算法

Seismic dip and azimuth parallel algorithm based on CUDA

为提高三维地震信号数据倾角、方位角算法的计算速度,针对其计算过程特点,提出了一种基于CUDA的并行算法。首先,对倾角方位角算法进行模块化分析并找到算法的计算瓶颈;然后,分别从差分计算、滤波以及后续的倾角方位角计算步骤入手,进行并行策略分析;最后,通过线程以及显存优化来提高算法的计算速度。通过实验验证,在大于50×50工区大小的地震数据中可以实现7.7以上的加速比,并能保证实际工程需求的计算精度。该并行算法对于三维地震数据的均值与中值滤波以及地震信号差分计算提出了一种可行的并行加速方案,可以广泛应用于地震信号处理领域中。

To improve the computational speed of 3D seismic signal dip and azimuth algorithm,for the characteristics of its computational process,a parallel algorithm based on CUDA was given.Firstly,dip and azimuth algorithm was modularly analyzed and the computational bottleneck of the algorithm was found.Then,from the difference calculation,filtering and subsequent dip and azimuth calculation steps,parallel strategy analysis was carried out.Finally,thread and graphics memory optimization were used to improve the calculation speed of the algorithm.It is verified by experiments that the acceleration ratio of 7.7 times or more can be achieved for seismic data larger than 50×50,and the calculation accuracy of actual engineering requirements can be guaranteed.In the parallel algorithm.a feasible parallel acceleration scheme was proposed for the mean and median filtering of 3D seismic data and the differential calculation of seismic signals,which can be widely used in the field of seismic signal processing.