Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Convolutional neural network accelerator and acceleration method

A convolutional neural network and accelerator technology, applied in the field of convolutional neural network accelerators and acceleration, can solve problems such as limited computing resources and storage resources, achieve low power consumption, satisfy performance, and improve utilization

Pending Publication Date: 2020-04-21
TIANJIN UNIV
View PDF3 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, the computing resources and storage resources of the FPGA platform are limited. Under the limited resources, realizing a high-performance and low-power neural network accelerator is a technical problem that needs to be solved at present.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Convolutional neural network accelerator and acceleration method
  • Convolutional neural network accelerator and acceleration method
  • Convolutional neural network accelerator and acceleration method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0030] Such as Figure 1~3As shown, a convolutional neural network accelerator includes a data input control module 1, an input buffer 2, a row buffer module 3, a matrix module 4, a convolution processing unit 5, a weight input module 6, an output buffer module 7, The bias module 8 and the pooling module 9, wherein the data input control module 1 is used to receive data from the memory or the pooling module 9, and store the received input data into the input buffer 2 in a preset order; The line buffer module 3 is used to splice the input data, output multiple lines of data, and then transmit it to the convolution processing unit 5; the matrix module 4 is used to splice the input data into an input matrix, and then transmit it to the convolution processing unit 5; the weight The input module 6 is used to cache the weights and pass the weights to the convolution processing unit 5 in a preset order; the convolution processing unit 5 is used to perform convolution operations on th...

Embodiment 2

[0042] like Figure 1~3 As shown, a convolutional neural network acceleration method includes the following steps:

[0043] Step 1, preset network acceleration parameters, receive data from the memory or pooling module 9, and store the received input data into the input buffer 2 in a preset order,

[0044] Step 2: After the weight input module 6 prepares the corresponding weight, judge whether it is a 1×1 convolution operation, if yes, the data input control module 1 outputs the data of multiple channels at one time, otherwise, the data input control module 1 presses Output the data of each channel in the preset order, and then through the line buffer module 3, the data input in a single line will be output in three lines at the same time;

[0045] Step 3: Concatenate the data into a 3×3 input matrix through the matrix module 4, then perform convolution operations on the data in the input matrix and the corresponding weights, and then perform channel-wise accumulation on the ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a convolutional neural network accelerator. The convolutional neural network accelerator comprises a data input control module, an input buffer, a row buffer module, a matrix module, a convolution processing unit, a weight input module, an output buffer module, a bias module and a pooling module. A multiplexing parallelism degree structure is adopted, the utilization rate of a multiplier and an adder in the accelerator is increased, and the requirements for high performance, low power consumption and high flexibility are met under limited resources. In addition, the invention also discloses a convolutional neural network acceleration method.

Description

technical field [0001] The invention belongs to the technical field of network accelerators, and in particular relates to a convolutional neural network accelerator and an acceleration method. Background technique [0002] Convolutional neural network is a feed-forward neural network and one of the representative algorithms of deep learning. Convolutional neural networks mainly rely on convolutional layers, pooling layers, and fully connected layers to process output data and perform efficient feature extraction. In recent years, with the continuous optimization of convolutional neural network algorithms, convolutional neural networks have developed rapidly and are widely used in computer vision, natural language processing and other fields. However, the implementation of convolutional neural networks relies on a large number of computing resources and storage resources, which brings many challenges to the implementation of convolutional neural networks. The central proces...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06N3/04G06N3/063
CPCG06N3/063G06N3/045Y02D10/00
Inventor 刘强曾成龙
Owner TIANJIN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com