Font Size: a A A

Research And Implementation Of Parallel Visualization Method Of Big Data Based On Cloud Platform

Posted on:2019-07-30Degree:MasterType:Thesis
Country:ChinaCandidate:S H ShiFull Text:PDF
GTID:2428330545456448Subject:Electronics and Communications Engineering
Abstract/Summary:PDF Full Text Request
With the development of science technology,all kinds of scientific computing data are increasing.It is an urgent need to visualize the data in order to obtain the important information in the data.However,with the increasing amount of data,the required computing performance of visualization is getting higher.The scheme of visual rendering with traditional single machine is often too low in computing and too long in time to meet the need of large data visualization in scientific computing.In view of the poor computing performance and long rendering time of single machine data rendering,parallel rendering based cloud platform is adopted in this paper.Parallel rendering can significantly improve rendering speed and shorten rendering time.And the cloud platform has many features that many different calculated performance node can be picked and the cluster can be extended anytime and anywhere.The rendering environment moved from the local to the cloud platform can be more convenient for parallel rendering,and greatly improving the efficiency of visualization and saving the drawing time.In this paper,two aspects of the work has been done:(1)using ParaView to draw parallel visualization on the local cluster,and rendering the four sets of data with single node,double node and four node respectively.The test results show that parallel rendering of clusters can significantly increase computing power and shorten rendering time.(2)Deploying ParaView in the cloud platform for parallel visualization,computing performance of the free nodes of the Amazon dynamic elastic cloud and other related data are tested in this paper.The results show that the Amazon elastic computing cloud nodes are equivalent to the medium performance notebook,which can achieve the basic scientific calculation.Then the cluster in cloud platform builds for ParaView rendering test,the results show that the cluster based on cloud platform can not only speed up the rendering speed obviously as the same as the local cluster,but also more flexible that can increase computing nodes in need to further acceleration.The experiment results show that the parallel visualization on cloud platform is a good solution,which can bring greatly facilitates to the scientific data visualization for scientific research worker.
Keywords/Search Tags:Cloud platform, ParaView, visualization, parallel rendering
PDF Full Text Request
Related items