Out-of-core rendering of datasets, i.e. rendering of dataset which do not fit into memory, requires special techniques in order to provide interactive framerates. This also applies to volume rendering of large datasets since GPU memory, despite growing over the last years, is rather limited. Thus, tile-based approaches are often utilized to optimize the transfer of the visible data to the GPU. In this project, you will implement and investigate different data transfer options to GPU memory. Main focus is the benchmarking and the identification of bottlenecks regarding the transfer. This project requires programming in C/C++ and OpenGL. Some background in GPU programming and volume rendering is advantageous.