I have several hundred image processing filters written in heavily templated C++ code that I need to convert to some kind of GPU processing, and I’ve decided OpenCL is probably the best choice for cross-platform/cross-device support.
But I have several architectural questions about how to break my filters down into kernels and the best way to deal with multiple “passes”, memory management, etc. Basically I need to work out a strategy for converting the way the current filters work over to the “GPU way”, and I want to get it “right” before I start.
So I’m looking for an expert OpenCL developer who would be willing to consult with us. If there are any interested consultants here please private message me with your credentials and lets talk!