二维叠前深度偏移的并行算法
|
摘要
王真理，杨长春，张洪审，张得顺，符基昕，匡兵，王建立．二维叠前深度偏移的并行算法．石油地球物理勘探，1997，32（6）：784～788，817鉴于PVM软件是一种几乎能够在所有流行的计算机体系结构上运行的并行环境，本文提出了单程序多数据流（SPMD）方式的二维叠前深度偏移的并行其法，即走时表计算采用炮点循环并行方式，成像过程采用炮检距并行算法。采用PVM并行环境提供的消息传递方式进行进程间通讯，在SGIPowerChallenge和IBMSP2两种并行机上实现了其二维叠前深度偏移的并行其法。测试结果表明：本文的二维叠前深度偏移并行化算法效率是比较高的，在SGIPowerChallenge机上，当CPU数为4时，测得的加速比达3．91，并行效率达到0．98；在IBMSP2机上，当书点数为3时，测得的加速比为2．94，效率达到0．98。
In view of the fact that PVM software is a parallel environment which works nearly in all existing computer systems, we advance the parallel algorithm which is for 2-D prestack depth migration and behaves in the form of single program and multiple data flows (i. e. travel time table is computed by using shot-point circular parallel way, and image process by using offset parallel method). The parallel al-gorithm for 2-D prestack depth migration has been achieved in two parallel comput-ers (SGI Power Challenge and IBM SP2) by applying message transfer mode to in-terprogram communication under PVM environment. The experiments drew the following conclusion: the parallel algorithm for 2-D prestack depth migratlon has quite high efficiency; when CPU number of SGI Power Challenge computer is 4, the speed-up ratio will be 3-91,and the parallel efficiency O. 98; when the node of IBM SP2 computer is 3,the speed-up ratio will be 2.94,and the parallel efficiency 0.98.
In view of the fact that PVM software is a parallel environment which works nearly in all existing computer systems, we advance the parallel algorithm which is for 2-D prestack depth migration and behaves in the form of single program and multiple data flows (i. e. travel time table is computed by using shot-point circular parallel way, and image process by using offset parallel method). The parallel al-gorithm for 2-D prestack depth migration has been achieved in two parallel comput-ers (SGI Power Challenge and IBM SP2) by applying message transfer mode to in-terprogram communication under PVM environment. The experiments drew the following conclusion: the parallel algorithm for 2-D prestack depth migratlon has quite high efficiency; when CPU number of SGI Power Challenge computer is 4, the speed-up ratio will be 3-91,and the parallel efficiency O. 98; when the node of IBM SP2 computer is 3,the speed-up ratio will be 2.94,and the parallel efficiency 0.98.
引文
1 Keho T H, Beydoun W B.Paraxial ray Kirchhoff migration.Geophysics,1988,53(12):1540～1546
2杨长春等.地震叠前深度偏移方法流程及应用.地球物理学报,1996,39(4):409～415
3徐升等.射线追踪的微变同格方法.地球物理学报,1996,39(1):97～102
4 Langan R T,Lerche I,Cutler R T. Tracing of rays through heterogeneous media:an accurate and efficientprocedure.Geophysics,1985,50(9):1456～1465
5陈景良.并行算法导论,石油工业出版社,1992
2杨长春等.地震叠前深度偏移方法流程及应用.地球物理学报,1996,39(4):409～415
3徐升等.射线追踪的微变同格方法.地球物理学报,1996,39(1):97～102
4 Langan R T,Lerche I,Cutler R T. Tracing of rays through heterogeneous media:an accurate and efficientprocedure.Geophysics,1985,50(9):1456～1465
5陈景良.并行算法导论,石油工业出版社,1992
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心