Modulight : A Framework for Efficient Dynamic Interactive Scientific Visualization

The interactive scientific visualization applications are based on heterogeneous codes to implement simulation or data processing, visualization and interaction parts. These different parts need to be precisely assembled to construct an efficient application running in interactive time. Component-based approach is a good paradigm to express this kind of applications. The interactive scientific visualization domain is now classically extended with visual analysis applications. In this case, some parts of the application need to be added or removed dynamically during its execution. In this paper, we describe a component-based approach dedicated to dynamic interactive scientific visualization applications. We propose a framework called Modulight which implements our approach using the MPI2 library and the optimized socket library OMQ. The performance of this framework is also analyzed from a real-life application of molecular dynamics.