摘要
现有的波动方程正演及偏移成像大多是采用MPI技术实现并行,这种计算方式在集群之类的分布式系统上效果很好,在单个计算节点受到多种限制,不能发挥多核处理器的最大效能。本研究将波动方程频率正演模拟过程中的大型矩阵方程求解分配给多个计算核心,利用多核心并行计算方法,多个频点的求解过程可以同步计算完毕,有效提高了多核处理器的利用效率和代码执行速度。
Most of the existing wave equation forward modeling and migration imaging codes adopt MPI technology to realize parallel computing.This computing method is very effective in distributed systems such as clusters,and the single computing node is limited by many kinds of limitations,which cannot give full play to the maximum efficiency of multi-core processors.In this paper,the solution of the large matrix equation in the forward modeling of wave equation frequency is assigned to multiple computing cores.By using the multi-core parallel computing method,the solution process of multiple frequency points can be calculated synchronously,which effectively improves the utilization efficiency and code execution speed of the multi-core processor.
作者
岳晓鹏
付亚运
陈康靖
Yue Xiaopeng;Fu Yayun;Chen Kangjing(School of Science,Xuchang University,Xuchang 461000,China)
出处
《科学技术创新》
2022年第35期181-184,共4页
Scientific and Technological Innovation
基金
河南省高等学校重点科研项目基础研究计划(21B170003)
河南省自然科学基金(222300420280)
河南省高校科技创新团队支持计划(23IRTSTHN018)。
关键词
VTI介质
频率域
多核并行
数值模拟
VTI medium
the frequency domain
multi-core parallel
numerical simulation
