I'm starting to explore the idea of using GPUs to achieve high-ish volume audio transcoding (as opposed to dedicated DSPs or CPU-based transcoding). I have some experience running CPU-based transcoding using Intel IPP and get reasonable performance from that, but I'm hoping to harness the power of the GPU to achieve more bang-for-buck.
Unsurprisingly most of the information I've found relates to using GPUs for video transcoding, rather than audio transcoding.
I'd be interested in hearing from anyone who's looked at GPUs for audio transcoding before. Does this idea have merit, or are there any technical reasons why it won't work?
On a more practical level, if I was to start prototyping this, am I correct in thinking that Intel OpenCL SDK would be the right Intel package to start with (as opposed to say the MediaSDK)?
Thanks in advance for anyone who takes the time to reply!