Gt4-based Parallel Graphics System Design And Realization

Grid technology is proposed as a new computing platform technology after computer and network technology developed to a certain stage, which is used to integrate the distributed, dynamic, heterogeneous resources in the wide area network, and achieve full sharing of kinds of resources. This integration does not need high hardware performance, only using multiple single compute of lower cost to simulate massive supercomputers which can be successful to achieve fast data calculation. In contrast, the current single computer can not meet graphics-draw user's requirements of more and more high quantity of hardware of the system, so people look to the grid parallel computing. This article discusses the concept of "computing resources sharing" in grid, and analyzes the theory and concepts of parallel computing, and on this basis to achieve a grid-based parallel graphics system.This thesis first gives an overview of grid computing, then introduces the grid mapping system components used in GT (Globus Tookit), which focuses on the architecture of GT and procedures and principles of Gram; following introduces the grid-based classification of parallel computing, parallel programming model and Java multithreading, parallel computing architecture and its advantages; lastly, clearly analyzes the model grid-based operating division GJS (Grid- based Job Submittion) and the two definitions of fractal images which system needs to draw, provides a theoretical support for the design and implementation of the system.After understanding the key technologies and components of grid parallel computing, the thesis adopts GJS model in parallel execution, designs and implements parallel graphics system based on GT4, including single-computer graphics module, parallel graphics module, the system parameter setting module and the running data statistics module. single-computer graphics module is directly calculated using the local computer drawing MandelBrot sets and Julia sets, and according to data calculated to draw graphics; parallel graphics module is used to draw graphics grid environment, the specific process is as follows: first computing task decomposition, and then use the Gram GT components parallel computing task scheduler will be to the grid service center, which issues computing tasks in the grid environment based on the number of grid nodes, and finally when it tests that the computing tasks of all nodes are completed, the local computer recycles calculation data of MandelBrot sets and Julia sets, and draws the graph based on these; setting the system parameter module controls the size of computing tasks by setting graphics drawing area of MandelBrot sets and Julia sets; data statistics module shows the statistical time required by single and parallel graphics drawing.