Message scheduling method based on shift register

Abstract

The invention relates to a message scheduling method based on a shift register. The message scheduling method is characterized by comprising the following steps: step S1, determining specific message fields that need to be compared according to a message processing requirement; step S2, determining the number L of historical messages that need to be compared according to a message processing pipeline structure; step S3, determining the series M of the shift register according to the specific fields and storage resources; step S4, according to an order-preserving requirement in pipeline processing, dividing the messages without the order-preserving requirement between each other into N classes, wherein each class of messages correspondingly realize one shift register to store messages to be scheduled; placing the messages with the order-preserving requirement in a same shift register; and executing a specific field comparison function through an additional combinatorial logic, and finishing first-level scheduling; and step 5, performing second-level scheduling in a first-level scheduling result, and selecting one message and sending the message to a message processing pipeline.

Description

technical field [0001] The invention belongs to the technical field of computer communication, and in particular relates to a message scheduling method based on a shift register. Background technique [0002] Message is the abstraction of information carrier in the field of computer and communication. According to different protocols, the information required by the system is carried in the message. In order to improve the information throughput rate, the pipeline mechanism is usually used to process the message. However, there may be conflicts among the messages of each pipeline level in the pipeline. When a conflict occurs, the pipeline must be suspended until the previous message that caused the conflict is processed. Taking the memory consistency protocol processing of a multi-channel server as an example, when the address of a protocol message entering the pipeline is the same as that of a message being processed, the protocol processing pipeline must be stopped, and...

AI Technical Summary

Problems solved by technology

[0006] First, the utilization rate of memory resources is low. Messages classified according to specific fields need to be stored in N memories. Once one of the memories is full, all writing of such messages must be stopped, even if other memories still have storage

[0007] Second, it cannot meet the order preservation requirements between different types of messages

The order of the packets has been disrupted when the packets are classified, and the order of sending them into the pipeline cannot be guaranteed

Images