Method for configuring core resources in multi-core system, multi-core system and management core

Abstract

The invention discloses a method of allocating core resources on a multi-core system, which includes: pre-select a core as a management core in the multi-core system, with other cores as the function cores; when the management core detects the function cores in a production line phase in overload condition, according to the load information sent by the function cores in all production line phases, select one function core as the supplementation core from the function cores with low load condition in production line phases and add the supplementation core to the production line phase with the overload function cores. The invention also discloses the multi-core system and the management core and realizes the core resources occupation adjustment in all production line phases, according to the business load trend of all production line phases in the multi-core system, in order to improve the business processing efficiency of the multi-core system.

Description

technical field [0001] The invention relates to the technical field of multi-core processors, in particular to a method for configuring core resources on a multi-core system, a multi-core system and a management core. Background technique [0002] The emergence of multi-core processors greatly improves the performance of traditional single-core processors. However, with the rapid development of multi-core processors and the continuous enhancement of multi-core integration, the development of multi-core software technology is relatively lagging behind. In order to give full play to the advantages of multi-core, the industry chooses to run an operating system on each core of the multi-core processor or run a single-threaded application program that does not require operating system support. Each core completes a specific task, and multiple cores work together in a pipelined manner. The system allocates different numbers of cores to each stage of the pipeline according to requi...

AI Technical Summary

Problems solved by technology

The system will run normally when the network traffic is normal, but when some unusual business traffic suddenly increases on the network, the system may have problems

For example: when a large number of IP fragmented packets suddenly appear on the network, because the IP fragmented packets need to be reassembled in the IP processing stage, the load of the core in the IP processing stage will increase sharply, and at the same time, due to the reorganization of the packets, the IP processing stage will transfer to TCP The packets transmitted in the processing stage are relatively reduced, so the load on the core in the TCP processing stage is reduced accordingly, and it may even be in an idle state for a long time. Similarly, the load on the core in the APP processing stage is also reduced, resulting in a bottleneck in the entire system in the IP processing stage. At the same time, the cores in the TCP and APP processing stages are not fully utilized, resulting in a waste of core resources

[0004] From the above, due to the prior art, once the multi-core system is started, the functions of each core will not change, so the number of cores occupied by each stage of the pipeline will not change, so when the core of a certain stage When the load suddenly increases, it is likely to cause a system bottleneck, which makes the business processing efficiency of the multi-core system low

Images