Multi-Source Least-square Reverse Time Migration Based on MPI And CUDA Hybrid Accelerating Algorithm
H. Xue and Y. Liu
Event name: 79th EAGE Conference and Exhibition 2017
Session: Least-square and Marchenko Imaging
Publication date: 12 June 2017
Info: Extended abstract, PDF ( 1.44Mb )
Price: € 20
Least-squares reverse-time migration (LSRTM) is able to produce images with fewer artefacts, higher resolution and more accurate amplitudes over conventional migration method. Because of these benefits, LSRTM attracts greater interest in recent years. However, the computational cost of LSRTM is much more expensive than traditional migration method. In order to reduce the calculation time, we propose a method to combine CPU and GPU together to accelerate the calculation of multi-source LSRTM using MPI and CUDA. This method and can parallel both in coarse-grain and fine-grain and make full use of the hardware facilities. Synthetic examples show that the method can greatly reduce the calculation time than the traditional LSRTM.