A method and system for allocating acceleration resources of fpga cloud platform

Abstract

The present invention provides an FPGA cloud platform acceleration resource allocation method and system. The present invention allocates and coordinates the acceleration card resources according to the delay between the user host and the FPGA acceleration card deployed in each network segment. When the user applies for When using FPGA, the FPGA accelerator card with the smallest delay between the FPGA resource pool and the host is allocated to the user, so as to realize the allocation of FPGA cloud platform acceleration resources; the cloud monitoring and management platform can be based on the geographical location of each FPGA board in the FPGA resource pool. Differently, the transmission delay between the network and the virtual machine is obtained, and the board with the smallest delay is allocated to each user. In addition, it can effectively prevent unauthorized users from accessing the acceleration resources in the resource pool at will, and protect the owner of the resource pool. effective rights. The invention not only effectively protects the FPGA acceleration card not authorized for use by the user, but also ensures the minimum network delay of the board card allocated to the user, achieves the optimal acceleration effect, and improves the user experience.

Description

technical field [0001] The invention relates to the technical field of data centers, in particular to a method and system for allocating acceleration resources of an FPGA cloud platform. Background technique [0002] With the rapid development of "cloud data intelligence", the volume of data is expanding rapidly, and the demand for computing in data centers is also rising rapidly. Various applications such as deep learning in video transcoding, image compression and decompression, and genetic testing have far exceeded the capabilities of traditional CPU processors. [0003] In order to make up for the gap between demand and performance, the industry has begun to improve processing performance through hardware acceleration, which is a heterogeneous computing method using dedicated coprocessors. With the continuous enhancement of FPGA processing capabilities, more and more data centers have begun to use FPGAs for acceleration. For example, the data centers of Microsoft, Amazo...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a method and system for allocating accelerated resources on an FPGA cloud platform, aiming to solve the problem of inconsistent network delay caused by a large number of FPGA accelerator cards in the prior art, resulting in poor acceleration effects, and to realize guaranteed allocation to users The network delay of the board card is the smallest, achieving the best acceleration effect and improving the user experience

Images