摘要
不同数据间同步互斥是影响多核处理器核间通信的重要因素。为提高不同数据节点间的通信效率,降低通信延迟,并减少数据传输的丢包现象,结合自适应随机早期检测(ARED)算法和二进制指数退避(BEB)算法,提出一种软硬件结合的同步互斥管理分配方式。通过Matlab建模分析两种算法的可行性,并移植到ZYNQ7000开发平台中进行仿真测试。测试结果表明,该通信机制具有较低的丢包率和延迟时间,提高了整体传输效率,降低了通信过程延迟等待时间。
Synchronous mutual exclusion between different data is an important factor affecting the communication between multi-core processor cores. In order to improve the communication efficiency between different data nodes, reduce the delay of communication, and reduce the loss of data transmission, combining Adaptive Random Early Detection(ARED) algorithm and Binary Exponential Backoff(BEB) algorithm, a management allocation method of synchronous mutual exclusion that was combining software and hardware was proposed. The feasibility of the two algorithms was analyzed by Matlab modeling, and the synchronous mutual exclusion management allocation method was transplanted to the ZYNQ7000 development platform for simulation testing. The test results showed that the communication mechanism had a lower packet loss rate and delayed time, which improved the efficiency of overall transmission and reduced the delayed time of communication process.
作者
林康
张玲
于宗光
陈振娇
薛海卫
LIN Kang;ZHANG Ling;YU Zongguang;CHEN Zhenjiao;XUE Haiwei(School of Internet of Things Engineering,Jiangnan Untversity,Wuri,Jiangsu 214122,P.R.China;The 58th Research Institute,China Electromics Technology Group Corp.,Wuri,Jiangsu 214035,P.R.China)
出处
《微电子学》
CAS
北大核心
2021年第2期225-229,共5页
Microelectronics
基金
国家自然科学基金资助项目(61704161)。
关键词
同步互斥
ARED算法
丢包率
延迟
synchronous mutual exclusion
ARED algorithm
packet loss rate
delay
