Encog does not currently support GPU processing, either OpenCL or CUDA. In the past both platforms were used with Encog on an experimental basis. GPU programming is not an area of expertise for me, and is not something that I have time to research for Encog. GPU processing is a very cool technology, and if you are working on a serious neural network project you SHOULD be using GPU processing. At this point I am much more interested in using GPU libraries than developing the technology from the ground up. Additionally, there are already very advanced/supported libraries that support this technology. I primarily use GPU processing through Google TensorFlow and Deeplearning4J on Amazon AWS for projects that I am involved with.
If you are interested in the C/C++ version of Encog that I experimented with for CUDA, you can find it here. I do not presently use OpenCL, but have heard good things about Ivan Vasilev’s Neural Networks Project.