Kernel operation optimization method, device and system

An optimization method and technology of an optimization device, applied in the field of deep neural networks, can solve problems such as difficulty in guaranteeing and poor generality, and achieve the effect of high computing efficiency

Pending Publication Date: 2021-07-13
BEIJING XIAOMI PINECONE ELECTRONICS CO LTD
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this solution requires parameter tuning to be completed in advance on all computing cores, which is difficult to guarantee, resulting in poor versatility of this solution

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
  • Kernel operation optimization method, device and system
  • Kernel operation optimization method, device and system
  • Kernel operation optimization method, device and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0095] Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with aspects of the invention as recited in the appended claims.

[0096] A set of general parameters can be set for the calculation methods of various computing kernels, and the general parameters can be used for any input size and device computing power. However, this method ignores the differences of different computing cores. Since different input information and operating device environments require specific parameters to maximize efficiency, the general p...

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 relates to a kernel operation optimization method, device and system. The method relates to a deep neural network technology, and solves the problem that a general tuning scheme cannot give consideration to both high performance and universality. The method comprises the steps that before a calculation task is executed, whether kernel execution parameters of all kernel functions are optimized kernel execution parameters or not is checked; under the condition that the kernel function of which the kernel execution parameters are not optimized exists, requesting the optimized kernel execution parameters of the kernel function from a far-end server; and receiving an optimization scheme of the kernel execution parameters fed back by the remote server, and obtaining the optimized kernel execution parameters according to the optimization scheme. The technical scheme provided by the invention is suitable for a multi-kernel parallel computing scene, a universal adjusting and optimizing scheme is realized, and high computing efficiency is ensured.

Description

technical field [0001] The present disclosure relates to deep neural network technology, and in particular to a kernel operation optimization method, device and system. Background technique [0002] As the computing power of chips on mobile phones and other devices becomes stronger and stronger, more and more applications will hand over part of the computing tasks to the device side, which can effectively save server resources while bringing better user experience. In recent years, the popularity of deep neural networks has promoted the development of device-side computing. The computing capabilities of various devices are improving rapidly, and various deep learning proprietary computing devices are also emerging in an endless stream. Due to the high threshold for the use of proprietary chips, the most widely used ones are still the central processing unit (CPU) and the image processing unit (GPU). Compared with CPU, GPU has great advantages in high-performance computing, ...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06F9/50
CPCG06F9/5027
Inventor 刘琦何亮亮
Owner BEIJING XIAOMI PINECONE ELECTRONICS CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap