GPU acceleration has great potential. In the past 5 years or so, GPUs have been widely used for non-graphics-related purposes. They are extremely useful for accelerating algorithms which are parallel, and not inherently serial. When I heard that LR was finally using GPU acceleration, I was excited. Unfortunately, Adobe is apparently only using OpenGL for rendering - the final stage of processing. I had hoped that they were using a modern parallel development model, maybe using OpenCL (which is a pain, but can give excellent, portable results), OpenACC (which is easy to use, but has limited support), CUDA (an excellent programming environment, but limited to Nvidia graphics cards), or OpenMP 4.0 (which seems to be the emerging standard - basically easy to use, and increasingly available).