Data processing method based on brain-like processor, router and network system

Abstract

The invention provides a data processing method based on a brain-like processor, a router and a network system, and the method comprises the following steps: obtaining a multicast routing table, and determining a multicast routing protocol in the multicast routing table; determining a data transmission route through a routing algorithm according to a multicast routing protocol; acquiring a data packet, and determining a transmission priority of the data packet through priority arbitration according to a target node of data packet transmission; outputting the data packet according to the output address of the data transmission route and the transmission priority; according to the scheme, the number of crossbar switch ports is reduced, the overall area of the router is reduced, a self-adaptive priority arbitration mechanism is added, link transmission fairness can be achieved, the method can be used for inter-core communication of a multi-core brain-like processor, the requirement of the multi-core brain-like processor for reducing cost including power consumption and area of inter-core communication is met, and the method is suitable for popularization and application. Meanwhile, high-performance requirements of low delay, high bandwidth and the like are provided, and the method can be widely applied to the technical field of network-on-chip.

Description

technical field [0001] The invention relates to the technical field of on-chip networks, in particular to a data processing method, router and network system based on a brain-like processor. Background technique [0002] With the advent of the fourth artificial intelligence revolution, brain-inspired science and artificial intelligence research have achieved remarkable development, and have been highly valued by researchers in academia and industry. In daily life, artificial intelligence has been widely used in many fields of technological innovation and daily life. Especially in complex tasks such as image classification, pattern recognition, and other complex tasks, brain-like processors have shown very good performance. At first, most brain-like processor designs were only done by academic research institutions, but as researchers demonstrated the great potential of these brain-like processors, the industry also began to pay attention to and invest in brain-like chip res...

AI Technical Summary

Problems solved by technology

[0004] However, in related technologies, in low-power embedded systems, the gap between complex computing requirements and the availability of hardware resources is getting bigger and bigger, and the key point of multi-core brain-like processors is to reduce the area and power consumption of the interconnection structure and Putting more neurons on a chip and how to achieve efficient mass data transfer

In related technologies, the communication network architecture based on multi-core brain-like processors has the following shortcomings or deficiencies. First, the scalability is poor, and only one master device can transmit data at the same time; As the number of nodes increases, the cost of hardware overhead also doubles; second, the existing communication network architecture involves complexity, and the cost of node area is large

Images