FPGA network for stream computing and stream computing system and method

Abstract

The invention relates to an FPGA (Field Programmable Gate Array) network for stream computing, the FPGA network is formed by communication connection of a plurality of FPGAs and is used for processinguser data to obtain a processing result, and each FPGA comprises a computing logic unit group module used for performing data processing to obtain an intermediate result; a control logic module usedfor transmitting the user data to the computing logic unit group module, transmitting the intermediate result to the next FPGA according to a routing table, or transmitting the intermediate result asa processing result to an FPGA network outlet; and a communication module used for carrying out external data transmission of the FPGA. According to the method, software and hardware cooperate with each other to work, parallel components in the big data calculation data flow can be fully mined, the high-parallelism calculation advantage of the FPGA is brought into play, the calculation delay of flow calculation is greatly reduced, and the feedback speed of big data application affairs such as a recommendation system and a calculation advertisement is increased.

Description

technical field [0001] The present invention relates to the technical field of big data computing, in particular to a flow computing acceleration method and system. Background technique [0002] Stream computing is a distributed big data computing model. In stream computing, the arrival time and order of data cannot be determined, nor can all these data be stored, but real-time analysis and processing of data is required. The stream computing mode requires the support of a stream computing framework (such as Apache Storm). An application is usually divided into multiple computing components and distributed among several servers to form a relatively fixed topology. Data flows between components through the network , in the era of big data, data is often presented in the form of streaming data. Big data applications such as computational advertising, public opinion analysis, and recommendation systems need to analyze and judge user behavior in real time during the user's use p...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to overcome the defects of high energy consumption and communication delay in flow computing in the above-mentioned prior art, and propose a flow computing system and method based on FPGA network, which is based on FPGA network and uses software and hardware to cooperate with each other to work , to reduce the computing delay of streaming computing and improve the feedback speed of big data application transactions

Images