基于多线程多GPU并行加速的最小二乘逆时偏移算法
详细信息    查看全文 | 推荐本文 |
  • 英文篇名:Least-squares reverse time migration based on multi-thread and multi-GPU parallel acceleration
  • 作者:柯璇 ; 石颖 ; 张伟 ; 张振 ; 何伟
  • 英文作者:KE Xuan;SHI Ying;ZHANG Wei;ZHANG Zhen;HE Wei;School of Earth Science,Northeast Petroleum University;Research Institute of Exploration & Development,PetroChina Tarim Oilfield Company;Nangfang Branch of Sinopec Geophysical Corporation;
  • 关键词:时域最小二乘逆时偏移 ; GPU ; 多线程 ; Pthread ; 存储器优化 ; 共享存储器 ; 寄存器
  • 英文关键词:time-domain least-squares reverse time migration;;GPU;;multi-thread;;Pthread;;memory optimization;;shared memory;;register
  • 中文刊名:SYWT
  • 英文刊名:Geophysical Prospecting for Petroleum
  • 机构:东北石油大学地球科学学院;中国石油塔里木油田分公司勘探开发研究院;中石化石油工程地球物理公司南方分公司;
  • 出版日期:2019-01-25
  • 出版单位:石油物探
  • 年:2019
  • 期:v.58
  • 基金:国家自然科学基金项目(41574117,41474118,41804133);; 黑龙江省杰出青年科学基金项目(JC2016006)共同资助~~
  • 语种:中文;
  • 页:SYWT201901011
  • 页数:15
  • CN:01
  • ISSN:32-1284/TE
  • 分类号:92-106
摘要
最小二乘逆时偏移算法可对地下复杂构造精确成像,但由于计算量大,目前仍难以在实际资料处理中广泛推广应用,因此研究该方法的高效计算策略具有重要意义。结合Pthread标准,提出了多线程多图形处理器(Graphics Processing Unit,GPU)并行加速策略,在共炮点道集域分解计算任务,由多GPU并行计算并实时更新数据;并结合GPU存储器优化方法,调用GPU端共享存储和寄存器等高速存储器,提高波场模拟的计算效率;最终实现了二维空间的时域最小二乘逆时偏移算法大幅加速计算。分别对Marmousi2截断模型和Marmousi模型进行加速成像测试,结果表明:基于多线程多GPU并行加速的最小二乘逆时偏移算法具有普适性;随着数据规模的增加,该方法的加速效率可逐渐逼近线性加速,数据同步延迟小,加速效率显著。
        Least-squares reverse time migration(LSRTM)can accurately obtain images of subsurface structures.However,its application to actual data can be challenging owing to the large amount of calculation required.Therefore,aparallel acceleration strategy that combines Pthread standard with multithread-driven multi-graphics-processing-unit(multi-GPU)is proposed.The method decomposes computing tasks in common-shot gather domain with real-time update of data through multi-GPU parallel computing.The GPU memory optimization method is used to invoke high-speed memory,such as shared memory and registers,to increase the computational efficiency of wave field modeling.Finally,acceleration of 2Dtime-domain LSRTM is realized.The method was tested using the Marmousi2 truncation and Marmousi synthetic data.The results showed that the proposed method is applicable to different types of data.With the increase in data scale,the acceleration efficiency can gradually approximate linear acceleration,and is enhanced with small data synchronization delay.
引文
[1]SCHNEIDER W A.Integral Formulation for migration in two and three dimensions[J].Geophysics,1978,43(1):49-76
    [2]CLAERBOUT J.Toward a unified theory of reflector mapping[J].Geophysics,1971,36(3):467-481
    [3]GAZDAG J.Wave equation migration with the phaseshift method[J].Geophysics,1978,43(7):1342-1351
    [4]HEMON C.Equations d’onde et modeles[J].Geophysical Prospecting,1978,26(4):790-821
    [5]BAYSAL E,KOSLOFF D D,SHERWOOD J W C.Reverse time migration[J].Geophysics,1983,48(11):1514-1524
    [6]MCMECHAN G A.Migration by extrapolation of timedependent boundary values[J].Geophysical Prospecting,1983,31(3):413-420
    [7]TARANTOLA A.Inversion of seismic reflection data in the acoustic approximation[J].Geophysics,1984,49(8):1259-1266
    [8]LAMBAREG,VIRIEUX J,MADARIAGA R,et al.Iterative asymptotic inversion in the acoustic approximation[J].Geophysics,1992,57(9):1138-1154
    [9]NEMETH T,WU C,SCHUSTER G T.Least-squares migration of incomplete reflection data[J].Geophysics,1999,64(1):208-221
    [10]郭书娟,马方正,段心标,等.最小二乘逆时偏移成像方法的实现与应用研究[J].石油物探,2015,54(3):301-308GUO S J,MA F Z,DUAN X B,et al.Research of leastsquares reverse-time migration imaging method and its application[J].Geophysical Prospecting for Petroleum,2015,54(3):301-308
    [11]YAO G,JAKUBOWICZ H.Least-Squares ReverseTime Migration[J].Expanded Abstracts of 82nd Annual Internat SEG Mtg,2012:3406-3410
    [12]郭振波,李振春.最小平方逆时偏移真振幅成像[J].石油地球物理勘探,2014,49(1):113-120GUO Z B,LI Z C.True-amplitude imaging based on least-squares reverse time migration[J].Oil Geophysical Prospecting,2014,49(1):113-120
    [13]ZHANG Y,DUAN L,XIE Y.A stable and practical implementation of least-squares reverse time migration[J].Geophysics,2015,80(1):V23-V31
    [14]ZHANG Q,ZHOU H,CHEN H,et al.Least-squares reverse time migration with and without source wavelet estimation[J].Journal of Applied Geophysics,2016,134(1):1-10
    [15]李庆洋,黄建平,李振春,等.去均值归一化互相关最小二乘逆时偏移及其应用[J].地球物理学报,2016,59(8):3006-3015LI Q Y,HUANG J P,LI Z C,et al.Mean-residual normalized cross-correlation least-squares reverse time migration and its application[J].Chinese Journal of Geophysics,2016,59(8):3006-3015
    [16]刘学建,刘伊克.表面多次波最小二乘逆时偏移成像[J].地球物理学报,2016,59(9):3354-3365LIU X J,LIU Y K.Least-squares reverse-time migration of surface-related multiples[J].Chinese Journal of Geophysics,2016,59(9):3354-3365
    [17]BERKHOUT A J.Areal shot record technology[J].Journal of Seismic Exploration,1992,1(3):251-264
    [18]黄建平,李闯,李庆洋,等.一种基于平面波静态编码的最小二乘逆时偏移方法[J].地球物理学报,2015,58(6):2046-2056HUANG J P,LI C,LI Q Y,et al.Least-squares reverse time migration with static plane-wave encoding[J].Chinese Journal of Geophysics,2015,58(6):2046-2056
    [19]LI C,HUANG J P,LI Z C,et al.Preconditioned prestack plane-wave least squares reverse time migration with singular spectrum constraint[J].Applied Geophysics,2017,14(1):73-86
    [20]DAI W,HUANG Y,SCHUSTER G T.Least-squares reverse time migration of marine data with frequencyselection encoding[J].Expanded Abstracts of 83rd Annual Internat SEG Mtg,2013:3231-3236
    [21]李闯,黄建平,李振春,等.平面波最小二乘逆时偏移编码策略分析[J].石油物探,2015,54(5):592-601LI C,HUANG J P,LI Z C,et al.Analysis on encoding strategies of plane-wave least-square reverse time migration[J].Geophysical Prospecting for Petroleum,2015,54(5):592-601
    [22]LIU Y S,TENG J W,XU T,et al.An efficient steplength formula for correlative least-squares reverse time migration[J].Geophysics,2016,81(4):S221-S238
    [23]李闯,黄建平,李振春,等.预条件最小二乘逆时偏移方法[J].石油地球物理勘探,2016,51(3):513-520LI C,HUANG J P,LI Z C,et al.Preconditioned leastsquares reverse time migration[J].Oil Geophysical Prospecting,2016,51(3):513-520
    [24]WU D,YAO G,CAO J,et al.Least-squares RTM with L1norm regularization[J].Journal of Geophysics and Engineering,2016,13(5):666-673
    [25]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):2938-2943LI B,LIU H W,LIU G F,et al.Computational strategy of seismic pre-stack reverse time migration on CPU/GPU[J].Chinese Journal of Geophysics,2010,53(12):2938-2943
    [26]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):1725-1733LIU H W,LI B,LIU H,et al.The algorithm of high order finite difference pre-stack reverse time migration and GPU implementation[J].Chinese Journal of Geophysics,2010,53(7):1725-1733
    [27]SHI Y,WANG Y H.Reverse time migration of 3Dvertical seismic profile data[J].Geophysics,2016,81(1):S31-S38
    [28]石颖,刘洪,邹振.基于波动方程表面多次波预测与自适应相减方法研究[J].地球物理学报,2010,53(7):1716-1724SHI Y,LIU H,ZOU Z.Surface-related multiples prediction based on wave equation and adaptive subtraction investigation[J].Chinese Journal of Geophysics,2010,53(7):1716-1724
    [29]石颖,王维红,李莹,等.基于波动方程三维表面多次波预测方法研究[J].地球物理学报,2013,56(6):2023-2032SHI Y,WANG W H,LI Y,et al.3Dsurface-related multiple prediction approach investigation based on wave equation[J].Chinese Journal of Geophysics,2013,56(6):2023-2032
    [30]郭雪豹,刘洪,石颖.基于频域衰减的时域全波形反演[J].地球物理学报,2016,59(10):3777-3787GUO X B,LIU H,SHI Y.Time domain full waveform inversion based on frequency attenuation[J].Chinese Journal of Geophysics,2016,59(10):3777-3787
    [31]CLAERBOUT J.Earth soundings analysis:processing versus inversion[M].Boston:Blackwell Scientific Publication,2004:131-153
    [32]YANG P L,GAO J H,WANG B L.Agraphics processing unit implementation of time-domain full-waveform inversion[J].Geophysics,2015,80(3):F31-F39
    [33]MICIKEVICIUS P.3Dfinite difference computation on GPUs using CUDA[C]∥Proceedings of 2nd Workshop on General Purpose Processing on Graphics Processing Units.New York:ACM,2008:79-84

© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号

地址:北京市海淀区学院路29号 邮编:100083

电话:办公室:(+86 10)66554848;文献借阅、咨询服务、科技查新:66554700