| Summary: | Modeling molecular surfaces enables us to extract useful information about interactions with other molecules, as well as measurements of molecular areas and volumes. Many types of algorithms have been developed to represent and rendering molecular surfaces. However, these algorithms have questionable time performance in the visualization of molecular surfaces because they are usually designed to run CPU. A possible solution to resolve this problem is the use of parallel computing, but parallel computing systems are in general very expensive. Fortunately, the appearance of the new generation of low-cost programmable GPUs with massive computational power can, in principle, solve this problem. So, in this thesis we present a GPU-based algorithm to speed up the rendering of molecular surfaces. Besides we carry out a study that compares a sequential version (CPU) to a parallel version (GPU) of well-know Marching Cubes (MC) algorithm to render Connolly surface, as well as van der Waals surfaces.
|
|---|