High-reliability link failure tolerance module and method aiming at transient failures and intermittent failures in network-on-chip

Abstract

The invention discloses a high-reliability link failure tolerance module and method aiming at transient failures and intermittent failures in network-on-chip. The module and method are characterized in that a separation type ECC is adopted to detect whether data is wrong in the network in real time and define transient failures and intermittent failures; a retransmission buffer area arranged in a router is used, and retransmission is conducted through backup data in the retransmission buffer area when transient failures occur in a link to make data go wrong and can not be corrected; a backup header flit in a virtual channel is used, and when intermittent failures occur in the link to make data go wrong and can not be corrected, data packet transmission is intercepted, and rerouting or resource release is conducted by adding header flit or tail flit to interceptive data again. With low hardware cost, network reliability can be effectively improved to guarantee system performance when failures occur.

Description

technical field [0001] The invention belongs to the technical field of fault-tolerant integrated circuit chip design, in particular to a high-reliability link fault-tolerant module and a method thereof for transient faults and intermittent faults in an on-chip network. Background technique [0002] With the development of semiconductor technology, more and more cores are integrated on a single chip. ) is proposed as a new on-chip multi-core system interconnect communication architecture solution due to its advantages of high scalability, low latency and high bandwidth. [0003] The main function of the NoC system is to ensure that the data packets can be transmitted correctly and losslessly from the source node to the destination node through the router. As a key data path connecting routers, links play a vital role. However, due to problems such as soft errors, crosstalk between lines, temperature and aging, link transmission reliability has been greatly challenged. When...

AI Technical Summary

Problems solved by technology

Due to the lack of tail flits (flit) to release the resources occupied by the data that has passed the faulty link, long-term resource occupation will cause network congestion and reduce network performance; similarly, due to the existence of the faulty link, The data of the faulty link lacks the routing guidance of the header flit, and occupying Buffer resources for a long time will cause network congestion and may even lead to deadlock

Images