Check patentability & draft patents in minutes with Patsnap Eureka AI!

A method and device for clustering based on cuda

A clustering and matrix multiplication technology, applied in the field of data processing, can solve problems such as inability to optimize CUDA processing, and achieve the effect of efficient operation

Active Publication Date: 2018-03-09
INSPUR BEIJING ELECTRONICS INFORMATION IND
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in the existing method of optimizing CUDA using the K-means clustering method, only regular matrices (referring to matrices whose matrix dimension is a power of 2, especially common matrices with dimensions of 32, 64, 128, etc.) are optimized. , for matrices with irregular dimensions, CUDA cannot be used for optimization, so that the K-means clustering method can run efficiently under CUDA

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
  • A method and device for clustering based on cuda
  • A method and device for clustering based on cuda
  • A method and device for clustering based on cuda

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0057] In this embodiment, it is assumed that there is an input file containing a point array of 100 points, and the dimension of the array file is 18 (that is, each point is represented by 18 numbers), and these point arrays need to be divided into 10 classes.

[0058] First, create a 100*18 two-dimensional array in the CPU as a CPU point array, and create a 10*18 two-dimensional array as a CPU class array, and read the CPU point array and CPU class array;

[0059] Similarly, a GPU point array and a GPU class array are established in the GPU memory. In order to clearly illustrate this embodiment, a clustering result array and a temporary array for storing GPU matrix operation results are correspondingly established according to the CUDA point array. Among them, each element of the clustering result array is initialized to -1. Here, the GPU point array is obtained by transposing the CPU point array, so when performing matrix calculations, relevant definitions are required.

...

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 present application discloses a method and device for clustering based on CUDA, including: dividing the determined Computing Unified Device Architecture (CUDA) class array and CUDA point array into at least one 16*16 matrix; For a matrix of 16, the assignment flag bit is 0, and the other bits are 1; for each divided matrix, perform matrix operations in a form similar to matrix multiplication; for matrices less than 16*16, multiply the flag bits when the results are accumulated, Obtain the result of the GPU matrix operation; count the results of the GPU matrix operation, and when the number of points changed in the matrix operation result is not less than the preset threshold, update the CUDA class array according to the matrix operation result until the clustering is completed. The present invention divides the determined CUDA class array and CUDA point array into a matrix of 16*16; for a matrix less than 16*16, the assignment flag is 0, and the other bits are 1. Obtain the result of GPU matrix operation. The clustering of irregular matrices is realized to run efficiently under CUDA.

Description

technical field [0001] The present application relates to data processing technology, in particular to a method and device for realizing clustering based on Computing Unified Device Architecture (CUDA). Background technique [0002] Cluster analysis, also known as group analysis, is a statistical analysis method for studying (samples or indicators) classification problems, and it is also an important algorithm for data mining. Cluster analysis is based on similarity, with more similarities between patterns within a cluster than patterns not within the same cluster. [0003] K-means algorithm is a hard clustering algorithm in cluster analysis, and it is a typical prototype-based objective function clustering method. The K-means algorithm uses a certain distance from the data point to the prototype as the optimized objective function, and uses the function The method of finding the extreme value obtains the adjustment rule of the iterative operation. The K-means algorithm us...

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 Patents(China)
IPC IPC(8): G06F17/30
CPCG06F18/23213
Inventor 沈铂张刚邱学侃胡金辉王娅娟张清
Owner INSPUR BEIJING ELECTRONICS INFORMATION IND
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com