摘要
为了解决现有面向可编程结构的资源开销过大且缺乏资源优化的问题,从解析器和匹配单元两个方面设计资源优化方法以及模型,展开优化前后资源开销对比分析。利用合并匹配域与类型域的方式来减小开销,采用协议状态转移图具体表示优化包头解析器的过程,通过选择最佳的流表宽度达到匹配单元资源优化的目的。分析结果表明:在包头域长度一定的条件下,优化处理资源开销比原先减小了近40%。当包头域长度相同时,经过优化后可以使资源开销下降近40%;随着处理单元数量的增加,还可以节省更多硬件资源。
In order to solve existing problems of excessive resource overhead and lackness of resource optimization for programmable architecture,the resource optimization method and model are designed from two aspects of parser and matching unit,and the resource cost before and after optimization are compared and analyzed.The cost is reduced by merging the matching field and the type field.The process of optimizing the header parser is represented by the protocol state transition graph,and the resource optimization of matching unit is achieved by selecting the best width of the flow table.The results show that when the length of the header field is fixed,the resource cost of optimized processing is reduced by nearly 40%.When the header field have same length,the resource cost can be reduced by nearly 40% after optimization.As the number of processing units increases,more hardware resources can be saved.
作者
赵雷
ZHAO Lei(Library,Shaanxi University of Traditional Chinese Medicine,Xianyang 712046,China)
出处
《自动化仪表》
CAS
2021年第9期51-53,60,共4页
Process Automation Instrumentation
关键词
网络数据
可编程
资源开销
硬件资源
资源优化
模型
包头域
元处理
Network data
Programmable
Resource cost
Hardware resources
Resource optimization
Model
Header domain
Meta processing
