Performance comparison of designated preprocessing white light interferometry algorithms on emerging multi- and many-core architectures

Parallel computing has been a niche for scientific research in academia for decades. However, as common industrial applications become more and more performance demanding and raising the clock frequency of conventional single-core systems is hardly an option due to reaching technological limitations, efficient use of multi-core CPUs has become imperative. 3D surface analysis of objects using the white light interferometry presents one of such computationally challenging applications. In this article three established preprocessing methods of white light interferometry data analysis are used to evaluate the suitability of three modern multi-core architectures - generic multi-core CPUs, GPGPUs and IBM's Cell BE. The results show that function offloading to GPGPUs, which offer independent memory and many hundreds of threads running in parallel, yields the highest performance compared to other systems. Furthermore, by outsourcing computational tasks to GPUs, the workload of other system resources, such as CPU or system memory, is reduced. This allows accelerated execution of other tasks, e. g. acquisition of images with higher frame rates.