Self-adaptive load balancing method for parallelization of spatial computation

A technology of adaptive load and space computing, applied in the field of high-performance computing, can solve the problems of reduced computing speed, high processor load, idle processor, etc., and achieve the effect of small communication load

Active Publication Date: 2011-05-18
TONGJI UNIV
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

These allocation and adjustment methods will cause some processors to be overloaded and some processors to be idle when the load distribution is uneven or changes. On the one hand, the calculation speed is reduced, and on the other hand, resources are wasted.

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
  • Self-adaptive load balancing method for parallelization of spatial computation
  • Self-adaptive load balancing method for parallelization of spatial computation
  • Self-adaptive load balancing method for parallelization of spatial computation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] The invention discloses a space division method oriented to space calculation, and dynamically adjusts the division line according to the load variation among multiple processors during the calculation process to achieve a new balance. Its characteristic is to accelerate the re-division of space by compressing or stretching the divided word space into groups.

[0044] In parallel space computing, each processor is responsible for computing a divided subspace, and the calculation amount comes from the number and complexity of the calculation objects in the space. It is advisable to set the calculation complexity of each calculation object to be the same, then the calculation amount The size of can be represented by the number of objects. During the calculation process, when the calculation object moves in the space, it is possible to migrate from the space responsible for one processor to the space responsible for another hammer, which causes the change of the calculatio...

Embodiment 2

[0070] Molecular motion simulation is an effective tool for revealing the laws of molecular motion and recreating physical and chemical reaction processes. However, the calculation amount of molecular motion simulation is huge, and it is difficult to realize effective simulation calculation with a single computer and serial methods. With the development of network and parallel computing technology, more and more problems that are difficult to solve due to the large amount of calculation have been solved. Parallel computing is to decompose heavy computing tasks into relatively small computing tasks, and then hand them over to multiple processors for simultaneous processing, with the purpose of reducing computing time as much as possible. In parallel computing, the distribution of tasks should be balanced as much as possible, so as to avoid the overweight tasks of some computing nodes and cause low overall computing efficiency, thus losing the meaning of parallel computing. For...

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 self-adaptive load balancing method for the parallelization of spatial computation, comprising the steps of: parallelizing spatial computation; in the process of computation, dynamically adjusting a dividing line according to load change conditions among a plurality of processors in order to reach new balance; and accelerating spatial re-division by compressing or extending the divided word spaces in a manner of grouping. During the parallelization of spatial computation, each processor is responsible for computing one divided subspace, and the computation quantity derives from the number and the complexity of the computed objects in the space; the computation complexities of the computed objects are set to be same, thus the computation quantity is represented bythe number of the objects; in the process of computation, when moving in the space, the computed objects are considered to move from the space for which one processor is responsible to the space for which another processor is responsible, which results in the change of computation load. The self-adaptive load balancing method provided by the invention can rapidly balance the loads among a plurality of processors responsible for the computation of subspaces, and simultaneously, small communication loads are realized among balanced hosts.

Description

technical field [0001] The invention belongs to the technical field of high-performance computing, and relates to a load balancing method, in particular to an adaptive load balancing method for spatial computing parallelization. Background technique [0002] At present, the research of network high-performance computing has been widely valued, and the construction ideas and implementation forms are also different. According to the form of construction, they can be divided into two categories related to technology: [0003] (1) Cluster computing. Use a high-speed network to connect a group of workstations or microcomputers to form a cluster, or find a group of idle processors on the general network to form a dynamic virtual cluster, so that it can provide high-performance computing services with high cost performance under the control of intermediate management. [0004] (2) Grid computing. Its goal is to organize and manage computing resources, data sources and other devi...

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): G06F9/46
Inventor 蒋昌俊张栋良陈闳中闫春钢丁志军张亚英
Owner TONGJI UNIV
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