Chimaera is not only specialized in finding solutions for specific image processing problems, we also keep the processing as fast as possible. Latest developments in computer graphics hardware technologies, like the introduction of Nvidia's CUDA or the standard OpenCL, allow to use the GPU (graphics processing unit) of any modern computer to act as a general purpose processing unit. Especially in image processing tasks, many algorithms can be parallelized and great speedups can be achieved by bringing the algorithms onto the GPU.
Chimaera has used hardware acceleration in algorithms for research and product development since the early days of general GPU processing. Therefore, our customers benefit from years of experience to take full advantage of these new technologies. We use GPU accelerated processing not only in our own algorithm developement, Chimaera also offers services for our customers to port existing algorithms to GPU languages. Of course, it makes a difference how these algorithms are parallelized and whether all concepts of the new technologies are optimally employed. In order to take the full benefit of the GPU's capability, we perform detailed performance analysis to draw conclusions about bottlenecks, better optimizations or the layout of the computing domain.