Linking Visualization and Scientific Understanding through Interactive Rendering of Large-Scale Data in Parallel Environment

Abstract

Finding the unknown laws of sciences among data is one of the most essential goals in modern scientific discoveries. Visualization can be a useful tool for domain scientists to understand data if these data can be interactive rendered. But interactive rendering and exploring of these massive, complex data sets has been identified as one of the major challenges in current scientific visualization field. This paper addressed the issue of real-time interactive rendering that fully utilizes the efficient and scalable heterogeneous parallel algorithms. We present an interactive parallel rendering framework for large-scale data visualization. Experiment results show that our schemes can offer stable and real-time interactive rendering ability even when visualizing 30 GB per time-step data with 32 nodes GPU.