Integration based Geometric Flow Visualization on the GPU

Praktikum/Diplomarbeit

Philipp Muigg

Content:

Description

The goal of this project is to extend an already existing Flow Visualization framework (SimVis, see www.simvis.at) by implementing integration based geometric flow visualization techniques such as stream lines, which follow the flow within one timestep, path lines, which are used to visualize time dependent flow fields and streak lines which mimic the injection of dye into the flow domain. In order to cope with large flow simulation datasets consisting of millions of cells the SimVis framework already supports decomposing the overall flow domain into individual chunks which can be handled indepently from each other. This adds additional complexity to the task of integrating geometric objects along the flow direction since they may cross between different chunks frequently. Thus the biggest challenge of this project will be the development of an efficient solution capable of rapidly integrating a large number of geometric objects along a flow field which is split into individual chunks. In order to achieve reasonable performance, parallelization will be crucial. Thus parallel programming languages/frameworks such as openMP, OpenGL in combination with GLSL, OpenCL and CUDA will play a key role when developing a fast AND scalable solution.

Task

Depending on whether this topic is chosen as Praktikum or as Master Thesis project the primary goals vary. In the former case the implementation of a parallel CPU based stream/path line generation method will be sufficient whereas the second option will require the implementation of a GPU based solution.

Requirements

Additional Information

Depending on the achieved results a certain financial renumeration is planned. Contact: Philipp Muigg
SimVis: www.simvis.at
Large parts of the programming should happen at the SimVis office which is located near the UNO city in the Tech Tower.