GPU Acceleration Of Micro-blog Diffusion Visualization

Graphics Processing Units(GPU),originally designed for graphics,texture and pixels rendering,now provide computational power for scientific applications.CUDA(NVIDIA's Compute Unified Device Architecture)is a general-purpose platform that supports existing program language.There is no memory to share between nodes in parallel programming and individual threads perform computations independently by using CUDA.So computation time and resources spent are reduced.In micro-blog platforms,message posting,forwarding and viewing are the basic activities of the social network.These elements all play important roles in information diffusion.Analyzing the information diffusion process can help reasoning about the trend of public opinions.When a large number of users participate in the information dissemination process,the user's behavior and interaction becomes highly complex.Many factors affect the diffusion of information,such as the importance of micro-blog contents,agreement and disagreement,the influence of fake news and hostile attacks by a spambot,(a program that posts a huge number of massages automatically).Using a computational fluid dynamic model,such as Lattice Boltzmann Method(LBM).LBM is a class of computational fluid dynamics(CFD)methods for fluid simulation.It has been proved a very effective method to visualizing such diffusion process by using a computational fluid dynamic model,such as LBM.By simulating streaming and collision processes across a limited number of particles,the intrinsic particle interactions evince a microcosm of viscous flow behavior applicable across the greater mass.However,it is difficult to visualize a large number of flow nodes by using LBM.This thesis presents an approach of LBM-based visualization by using GPU acceleration.Information diffusion by social network occurs when a large number of users are involved in the process.Fluid dynamic model has been proved to be an effective method to visualize such information.Fluid dynamic model is extremely timeconsuming for large scale computing.In this thesis,we use the CUDA toolkit to improve the fluid dynamic model(FDM)for higher efficiency.With GPU acceleration,the FDM approach provides a real-time visualization.