论文部分内容阅读
体系作战能力的发挥依靠军事通信网络物联的作战实体,而全面细致高效的日志分析是作战实体安全可靠运行的重要保障。本文针对日志分析过程中计算任务量大、分析效率低、开发周期长的问题,设计并实现了一套基于CUDA的大规模军事通信网络日志数据并行分析框架。单节点在处理日志数据分析的效率与传统CPU串行分析相比,可达到15~20倍的加速比。
The operational capability of the system relies on the operational entities of the military communications network IoT, and comprehensive and meticulous log analysis is an important guarantee for the safe and reliable operation of combat entities. In this paper, aiming at the problem of large computational workload, low analysis efficiency and long development cycle in log analysis, a set of CUDA-based parallel log data analysis framework for military communications network is designed and implemented. Single node in processing log data analysis efficiency compared with traditional CPU serial analysis, up to 15 to 20 times the speedup.