交叉节点缓存crossbar交换结构设计

摘要

The traditional exchange structure is complicated and time scheduling overhead problem. Using a modified round-robin scheduling algorithm, at the output buffer is set according to a certain order of output, and by verilog code to achieve the 8 * 8 CICQ switch fabric. Greatly reducing the conflict of input and output ports of traditional crossbar switch fabric, effectively avoiding HOL blocking problem . Using an algorithm complexity is O ( 1 ) of the round-robin scheduling algorithm , hardware implementation simple. It can achieve 100% throughput efficiency, to achieve the fastest speed three clock cycles low latency switching.%针对传统交换结构调度复杂且时间开销大的问题,采用交叉点缓存(Buffered Crossbar)交换结构和改进的轮询调度算法,在输出端设置按一定的缓存顺序输出,并通过verilog代码实现了8*8的CICQ交换结构.极大地缓解了传统Crossbar交换结构存在的输入输出端口冲突问题,有效避免了队头阻塞问题.采用算法复杂度为O(1)的轮询调度算法,硬件实现简单.可以达到100%的吞吐效率,实现了最快3个时钟周期的高速度低延时交换.