Because of low efficiency, seismic forward simulation of 3D wave equation by finite difference method cannot be applied to real large-scale product. To deal with this problem, the authors present a parallel accelerating solution based on multi-GPUs heterogeneous cluster. By using CPU/GPU cooperation parallel mode and domain decomposition method, the authors carry out a multi-level parallel strategy of this algorithm. Furthermore, non-blocking MPI communications, storage optimization and parallel I/O mechanism using MPI-IO API are presented to optimize computation efficiency and memory usage. This parallel 3D forward algorithm can effectively support large-scale practical production.
[1] Xia F,Dong L G,Mz Z T.The numerical modeling of 3-D elastic wave equation using a high-order,staggered-grid,finite difference scheme[J].Applied Geophysics,2004,1(1): 38-41.[2] 张文生,宋海斌.三维正交各向异性介质三分量高精度有限差分正演模拟[J].石油地球物理勘探,2001,36(4):422-432.[3] 杨仁虎,常旭,刘伊克.基于非均匀各向同性介质的黏弹性波正演数值模拟[J].地球物理学报,2009,52(9):2321-2327.[4] 王延光,匡斌.起伏地表叠前逆时深度偏移与并行实现[J].石油地球物理勘探,2012,47(2):266-273.[5] 刘红伟,李博,刘洪,等.地震叠前逆时偏移高阶有限差分算法及GPU实现[J].地球物理学报,2010,53(7):1725-1733.[6] 李博,刘红伟,刘国峰,等.地震叠前逆时偏移算法的CPU/GPU实施对策[J].地球物理学报,2010,53(12):2938-2943.[7] Michéa D,Komatitsch D.Accelerating a 3D finite-difference wave propagation code using GPU graphics cards[J].Geophys J Int,2010,182(1):389-402.[8] Komatitsch D,Erlebacher G,Gddeke D,et al.High-order finite-element seismic wave propagation modeling with MPI on a large GPUcluster[J].J Comput Phys,2010,229(20):7692–7714.[9] 龙桂华,赵宇波,李小凡,等.三维交错网格有限差分地震波模拟的GPU集群实现[J].地球物理学进展,2011,26(6):1938-1949.[10] 刘守伟,王华忠,陈生昌,等.三维逆时偏移GPU/CPU机群实现方案研究[J].地球物理学报,2013,56(10):3487-3496.[11] Roden J A,Gedney S D.Convolution PML (CPML):An efficient FDTD implementation of the CFS-PML for arbitrary media[J].Microwave and Optical Technology Letters,2000,27:334–339.[12] Toivanen J I,Stefanski T P,Kuster N,et al.Comparison of CPML implementations for the GPU-accelerated FDTD solver[J].Progress In Electromagnetics Research M,2011,19:61-75.