首页> 外国专利> METHOD AND DEVICE FOR PARALLEL ENUMERATION OF SUPPORT POINTS BASED ON METRIC SPACE

METHOD AND DEVICE FOR PARALLEL ENUMERATION OF SUPPORT POINTS BASED ON METRIC SPACE

机译：基于度量空间的支持点并行枚举的方法和装置

页面导航

摘要
著录项
相似文献

摘要

Disclosed in the present invention is a method and device for parallel enumeration of support points based on a metric space. The method comprises: calling a message delivery interface, and distributing data to be calculated to M nodes, one data set being assigned to each node; controlling the M nodes to respectively call a fork function, creating N sub-processes in each node, and assigning the data sets corresponding to the nodes to the N sub-processes, so that one sub data set is assigned to each sub-process, wherein N is equal to the number of GPUs comprised in each node; and controlling the N sub-processes created in each of the M nodes to operate, performing one-to-one control on the N GPUs in the corresponding node by means of the N sub-processes, using the GPUs to traverse data in the sub data sets of the corresponding sub-processes, carrying out parallel enumeration of support points, and storing the enumeration result. By means of the method above, the parallel enumeration of support points can be effectively implemented; the time consumed by enumeration of the support points is effectively shortened; and the performance of data calculation is improved.

机译：本发明公开了一种用于基于度量空间并行枚举支持点的方法和设备。该方法包括：调用消息传递接口，并将要计算的数据分配给M个节点，每个节点分配一个数据集。控制M个节点分别调用fork函数，在每个节点中创建N个子流程，并将与节点相对应的数据集分配给N个子流程，从而为每个子流程分配一个子数据集，其中，N等于每个节点中包括的GPU的数量;控制在所述M个节点中的每个节点中创建的N个子进程进行操作，借助于N个子进程对相应节点中的N个GPU进行一对一控制，并利用GPU遍历子节点中的数据相应子过程的数据集，对支持点进行并行枚举，并存储枚举结果。通过上述方法，可以有效地实现支撑点的并行枚举。有效地缩短了枚举支持点所花费的时间;并提高了数据计算性能。

著录项

公开/公告号WO2019227343A1

专利类型
公开/公告日2019-12-05

原文格式PDF
申请/专利权人 SHENZHEN UNIVERSITY;
展开▼

申请/专利号WO2018CN89041
发明设计人 MAO RUI;LIAO HAO;ZHOU CHI;HU ZILIANG;LIU KAINAN;LU MINHUA;LU KEZHONG;LUO QIUMING;LEI HAIJUN;CAI YE;WANG YI;
展开▼

申请日2018-05-30
分类号G06F19;
国家 WO
入库时间 2022-08-21 11:14:32

相似文献

专利
外文文献
中文文献