Research On Performance Optimization Method Of Large-Scale Graph Data Visualization System

In the era of big data,how to effectively display the results of big data processing is of great significance.The current Web visualization system for scale graph data processing needs to consume a lot of CPU resources to complete the calculation and rendering.When displaying a large amount of graph data,it is limited by browser performance,resulting in reduced efficiency.Many tool libraries have greatly improved the layout performance through continuous optimization of the algorithm,but the rendering is still inefficient.It is more suitable for processing large-scale operations on the CPU and Graphics Processing Unit(GPU),but the open-source tool library on the GPU only supports graphics rendering and lacks support for layout algorithms.In order to solve the above problems,a graph data visualization system combining CPU and GPU is designed and implemented.First,design and implement an efficient force-oriented layout(force-oriented layout)algorithm on the CPU,optimize the layout calculation process,and stop the layout after the system reaches initial stability in the layout;then,transfer the position information map data to the GPU,use the tool library that encapsulates the Web graphics library to design and implement the graphical representation of the data and complete the graphics rendering on the screen.Finally,monitor mouse events,complete various three-dimensional object objects,and complete user interaction by performing various transformations on-screen coordinates.Conversion,design background support system,provide graph data storage,and query interface of the computing system,pre-process the data.To confirm the efficiency of the system,the Wiki Talk data set in the real data set SNAP(Stanford Large Network Dataset Collection)was used for testing.Experiments prove that the system supports the query and visualization functions of the three application scenarios of N-degree relationship,relationship graph,and trusted relationship;for hundreds of nodes and even larger data sets,compared to traditional Graph Exp,GPU-based visualization System performance can be improved by about 10 times.