Modul

The student will gain an overview about the current parallel hardware architectures, which can be used for solving image processing problems. They will be able to analyse, parallelize and optimize image processing algorithms with respect to different hardware platforms. The introduction into OpenCL will give them the ability to implement the algorithms on graphic cards and multi-core processors.

The lecture will give an overview about the different forms of parallelism of an algorithm and the corresponding mapping on hardware architectures. It will introduce the different layouts of the hardware architectures and the different programming models.

To allow for a common programming model for both graphic cards and multi-core processors, the introduction into OpenCL will be very detailed. The OpenCL programming model, the API and the OpenCL C language will be explained. For an optimized use of OpenCL it is mandatory to understand the underlying memory model in all its details. With tutorials during the lecture the, students will get the opportunity to implement image processing algorithms on different architectures and optimize them accordingly. A graphic card and a Xeon-Phi accelerator board will be provided.