基于多图形处理单元加速的各向异性弹性波正演模拟  

Anisotropic Elastic Wave Forward Modeling Based on Multiple Graphics Processing Unit

作　　者：付小波[1] 马中高[2] 余嘉顺[1] 原健龙 韩超[1] FU Xiao-bo;MA Zhong-gao;YU Jia-shun;YUAN Jian-long;HAN Chao(College of Geophysics,Chengdu University of Technology,Chengdu 610059,China;Sinopec Geophysics Research Institute,Sinopec Key Laboratory of Geophysics,Nanjing 211103,China)

机构地区：[1]成都理工大学地球物理学院,成都610059 [2]中国石化石油物探技术研究院,中国石化地球物理重点实验室,南京211103

出　　处：《科学技术与工程》2018年第11期16-22,共7页Science Technology and Engineering

基　　金：中国石化地球物理重点实验室开放基金(wtyjy-wx2016-04-11)资助

摘　　要：比较分析了在不同网格大小介质模型情况下,分别采用串行计算、CPU 16个线程并行计算和4块图形处理单元(GPU)并行计算进行各向异性弹性波动方程正演模拟的执行时间差异。发现在网格点为256~3的大模型上,用4块GPU的并行模拟计算相对16线程并行计算与串行计算的加速分别为30倍与156倍。表明多GPU并行算法可以显著缩短数值模拟时间,而且模型网格越大,加速效果越显著。因此,在单机环境下进行大尺度模型的各向异性弹性波正演模拟,采用多GPU并行计算方式是一个合适的加速选择。Comparative analysis study of the computational efficiency of anisotropic elastic wave modeling using CPU serial calculation,graphics processing unit(CPU)16 threads parallel computing and multiple GPU parallel computing were performed on models of different size.The results show that compare to CPU serial calculation and CPU 16 threads parallel computing,multiple GPU parallel computing achieves a speedup of 30 and 156 times,respectively,on a model mesh of 256 3.Multiple GPU parallel computing can significantly speed up the modeling,and the speedup becomes more significant when the mesh is larger.Multiple GPU parallel computing is a proper solution for anisotropic elastic wave modeling on single node computing environment.

关 键 词：各向异性弹性波 有限差分 正演模拟 OpenMP/GPU并行计算 

分 类 号：P631[天文地球—地质矿产勘探]