面向复杂网络应用的CPU/FPGA异构网络处理系统协同性研究

With the expansion of network scale and the emergence of new applications and protocols, currently improving the processing performance and flexibility of network equipment has become the main issue to be solved urgently. The heterogeneous network processing system based on general purpose multi-core CPU with coprocessor provides a feasible technology method. However, this heterogeneous network processing system is beset by resource fusion, cooperation of hardware and software, capability reconfiguration, and so on. This project is aimed at complex network applications, and deeply investigates the key technologies of CPU/FPGA heterogeneous network processing architecture, mainly including architecture and development model of CPU/FPGA heterogeneous network processing system, optimization technology of basic framework performance and programmable packet forwarding acceleration technology supporting capability reconfiguration to enhance network processing performance. The research carries out intensive study of high-performance multi-core processor, network architecture, node architecture and network dedicated co-processor, which can provide a solid theoretical and technical basis for the design of new network processing architecture and hold great significance and application value for shortening development period of network processing platform and reducing development cost.

随着网络规模流量的扩大和新型应用及协议的涌现，提高网络设备的处理能力和灵活性成为当前亟待解决的主要问题。基于高性能通用多核CPU加协处理器的异构网络处理系统提供了一条可行的技术路径，然而这种异构网络处理系统受到资源融合性、软硬件协同性、能力重构性等问题的困扰。本课题面向复杂网络应用，深入研究通用多核CPU/FPGA异构网络分组处理架构协同性及加速机理实现关键技术，主要包括：资源融合的CPU/FPGA异构网络协同处理体系架构与开发模型、软硬件协同的异构网络处理系统基础框架优化技术、支持能力重构的可编程分组转发加速技术。本课题的研究可对高性能多核处理器、网络体系架构、网络专用协处理器、节点体系结构等研究成果充分挖潜，对新型网络处理平台实现研究积淀理论和技术基础，对缩短网络处理平台研制周期，降低研制成本具有重要意义和应用价值。

为了解决异构网络处理系统在资源融合性、软硬件协同性、能力重构性方面的问题，基于高性能通用多核CPU加协处理器的异构网络处理系统提供了一条可行的技术路径。本课题面向复杂网络应用，深入研究了通用多核CPU/FPGA异构网络分组处理架构协同性及加速机理实现关键技术，主要包括：资源融合的CPU/FPGA异构网络协同处理体系架构与开发模型、软硬件协同的异构网络处理系统基础框架优化技术、支持能力重构的可编程分组转发加速技术，并取得了理论成果和实用成果。本课题的研究可对高性能多核处理器、网络体系架构、网络专用协处理器、节点体系结构等研究成果充分挖潜，对新型网络处理平台实现研究积淀理论和技术基础，对缩短网络处理平台研制周期，降低研制成本具有重要意义和应用价值。