Path calculation method for optical on-chip network under optical circuit switching condition

Abstract

The invention discloses an optical chip on-chip network path calculation method under the condition of optical circuit switching, which is used to solve the problems of inability to perceive network flow and delay status, high network delay and low throughput in the existing method. The implementation steps are: initialize the set of candidate output ports avail_port; determine the number of candidate ports in the set and their idle status; when both candidate ports are occupied, calculate the predicted value TP of the processing delay of the two ports 1 、TP 2 and the congestion difference rate k; select the output port according to the comparison between k and the congestion difference rate threshold value K; when only one of the two candidate ports is idle, select the idle port as the output port; when both ports are idle, select the output port according to the above One-hop transmission direction selects the output port; when there is only one port, it is designated as the output port. The invention adopts congestion prediction to make link building groups consider network traffic and congestion when selecting a path, and reduce communication loss by the number of path turns, which can be used to realize congestion perception, reduce network delay, and improve communication reliability and throughput.

Description

technical field [0001] The invention belongs to the field of communication technology, and relates to an optical on-chip network path calculation method, in particular to an optical on-chip network path calculation method under the condition of optical circuit switching, which can be used for the path selection of data transmission between IP cores on the chip, and realizes the network path calculation method based on optical circuits. High-efficiency data transmission between IP cores in a circuit-switched optical network-on-chip, while maintaining low-latency and low-loss data communication. Background technique [0002] With the increasing demand for business communication in the information field, the number of IP cores on the processor chip is also increasing. The traditional network-on-chip based on electrical interconnection can no longer meet the communication requirements of the current network-on-chip in terms of energy consumption, delay, bandwidth, and crosstalk....

AI Technical Summary

Problems solved by technology

This path calculation method is only related to the location of the source and destination IP cores, and can only calculate a certain path for each packet. This method cannot perceive the traffic, congestion, and delay of the network, so that the path selected by the network for the packet cannot be optimal. Optimize the use of network resources, causing conflicts and congestion in network communication, and reducing delay and throughput performance

Therefore, as the number of IP cores in the chip continues to increase, the conflicts and congestions caused by inter-core communication based on optical circuit switching become very serious, and the use of dimensional order routing leads to higher communication delays and lower throughput.

Images