GPU-based Parallel Algorithm For Rendering Interactive Dynamic Water-surface

Since the invention of the computer, the computer technology maintains themomentum of rapid development. Not only the speed of the computer become fasterand faster, the volume of the computer become smaller and smaller, but also the eachsubdiscipline of computer develops very quickly, which also includes computergraphics and the emerging technology based on it–the virtual reality technology. Theemergence of virtual reality technology provides us with a new way of understandingthe world, changing the way people observe the world in the past.Water is an indispensable part of people’s daily life. In the virtual scene containswater, dynamic water becomes the focus of the picture, so the effect of watersimulation largely affects the overall effect of the scene. The simulation of water hasalways been a problem. Water has many special natures which determine thedifficulty of the water simulation. Firstly, water is a liquid, it has no fixed shape, itsmovement is subject to several uncertainties and there is almost nothing regularity;Secondly, reflection, refraction and other optical phenomena and water quality, waterdepth, and so on affect the color of the water; Thirdly, the water flow all the time,which requires each frame to be updated; Finally, it is necessary to sacrifice the effectof the screen to achieve real-time simulation of the water surface, otherwise, ordinaryPC is unable to render smooth screen under a huge amount of computing.In recent years, with the rapid development of graphic processing unit (GPU), itsfunction is no longer confined to graphic processing, but to enter into the field ofgeneral-purpose computing, so, more and more software use it to accelerate now. Thespeed advantage of the GPU parallel computing launches a strong impact on the statusof the central processing unit (CPU), especially after NVIDIA releases CUDAgeneral-purpose parallel computing architecture, which makes programming moreconvenient, more flexible, and easier to spread. With the help of GPU, we can put alarge number of complex computing tasks to GPU to alleviate the pressure of the CPU and accelerate the speed of program running.The main content of this paper is the real-time simulation of sea surface. The seasurface rendering algorithm uses the projected grid technology. We accelerated thecalculation of vertex coordinates of projected grid using the CUDA parallelcomputing technology, compared the speed of calculating a large amount of datausing CPU to calculating a large amount of data using GPU. Finally, the programimproved rendering speed significantly, so that the program can generate a morerealistic picture, while maintaining a satisfactory frame rate.