摘要： 提出一种利用图形处理单元(GPU)协同CPU的实时数据流分析匹配机制，将海洋实时数据流与国家标准数据进行比对，分析水质是否异常。该方法将存储到分布式基站上的数据分类规整并阶段化存储，接受数据流匹配请求后，在实时数据流匹配分析系统中利用GPU进行匹配分析处理，并自动生成分析结果。实验结果表明，该方法可以缩短海洋监测的检测时间并提高检测效率。

关键词: 图形处理单元, 数据流匹配, 数据流分析, 数字海洋, 海洋监测, 协同处理

Abstract: This paper presents a new method with Graphics Processing Unit(GPU) and CPU collaborate for matching the real-time marine data flow with the national standard data, and the result shows whether the water quality becomes abnormal. The method arranges and restores the marine data from distribute databases into a local stage of time series database, when system receives the requirement of matching, the GPU and CPU collaborate computing the marine data under the control of collaborate algorithm. The advantage of this method is both high-density computing and high bandwidth and the system can auto generate the matching result. Experimental results show that the method can effectively shorten the test time and improve detection efficiency.

Key words: Graphics Processing Unit(GPU), data flow matching, data flow analysis, digital ocean, ocean monitor, collaborative processing

中图分类号: 

• TP311