Network Processor Parallel Architecture And Performance Improvements

Nowadays, network processor (NP) are considered as a key technology that promotes the development of next generation network, and it has caught people's attention more and more. The appearance of network processor is thought to be a revolution of the communication industry. The market has accepted the contribution to network development of NP. Many companies and universities put people in the research of NP.This paper, first, studies the parallel architecture of NP, and describe the chief design method and parallel processing technology of NP from hardware and software, which gives the profile of NP's parallel architecture. Especially, this paper analyze the common software design model of NP, they are task pipeline model and thread pool model, and gives the advantages and disadvantages individually.Then, This paper presents a system-level methodology that consists of a programming model and optimization techniques for solving the functionality and memory-mapping problem on NP. The proposed techniques are evaluated by implementing three representative NP applications on the Intel IXP2400 processor, which belongs to the class of NP based multi-threaded architectures. This methodology will enable designers to map network-processing applications onto the current NP architectures without the time consuming process of manually optimizing through experimentation. It will accelerate the design flow for products, leading to shorter design turn around time.At last, this paper studies another important issues in NP, which is the local resource schedule optimization, and presents a new multi-processors schedule optimization algorithm, named Forecasting-Based Load Balance Hash Dispatch Algorithm (FLBHDA). This algorithm performs well on the load balance, low delay and packet ordering.