AMD and Cyberlink announced that the companies are expanding the current "engineering engagement with a strategic focus on Microsoft DirectX 11 DirectCompute", aiming to accelerate the computationally intensive tasks in Cyberlink's app suite.

The tasks in question are transcoding, automated facial recognition and tagging, video editing and processing with ATI Stream technology, all the while harnessing the power of the upcoming Win 7's DirectCompute API.

It seems like CUDA really scared the bejesus out of AMD as their initially meek ATI Stream now seems to be one of the focal points of development, and GPGPU processing might soon fully live up to the expectations.

You probably know this, but it's well worth repeating - the goal is to use the GPU to take some strain off of the CPU and harness its power for more intensive tasks, resulting in shared parallel processing between the GPU and CPU. This means that some computations once handled by the CPU, will be handled by the GPU allowing for much quicker transcoding and similar computationally-intensive tasks.