FPGA virtualization method for cloud deep learning reasoning

Abstract

The invention discloses an FPGA virtualization method for cloud deep learning reasoning, and the method comprises the following steps: introducing a two-stage instruction scheduler and a hardware resource pool to a deep neural network accelerator architecture based on an instruction set architecture, and constructing the deep neural network accelerator virtualization architecture based on the instruction set architecture; dividing a complete compiling process into static compiling and dynamic compiling according to a deep neural network accelerator virtualization architecture based on an instruction set architecture; generating a fine-grained instruction packet when static compiling is deployed for the first time, and when dynamic compiling is reconfigured, integrating the fine-grained instruction packet to quickly generate a demand instruction file. Rapid dynamic reconfiguration during operation is realized under the condition of ensuring multitask computing power resource sharing ofmaximizing the utilization rate of hardware resources.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence virtualization, in particular to an FPGA virtualization method for cloud deep learning reasoning. Background technique [0002] Currently in the era of rapid development of artificial intelligence, deep learning is playing an increasingly important role in various fields. Among them, the inference task of deep neural network (DNN) occupies most of the deep learning task load of cloud data center. The use of traditional general-purpose processors (Central Processing Units) in data centers can no longer meet the huge computing power requirements of deep learning. Therefore, dedicated hardware platforms such as GPUs, FPGAs (Field Programmable Gate Arrays) and ASICs (Application Specific Integrated Circuits) are now commonly used to accelerate deep learning algorithms. Thanks to FPGA's good balance of programmability, performance, and power consumption, more and more cloud se...

AI Technical Summary

Problems solved by technology

Therefore, the third option cannot ensure maximum resource utilization in any case

[0005] Based on this, there is an urgent need for a method to solve the problems of low resource utilization and high reconfiguration overhead in FPGA virtualization solutions for deep learning inference applications

Images