I am a professional visual effects software engineer in the film industry, and personally, I’d say that GPU audio is probably one of the most oversold pieces of technology currently out there.
GPU’s have, by design, relatively poor latency, especially as complexity increases. So low latency audio computation on GPU’s is going inherently against the way they were designed.
We use GPU’s for some aspects of rendering (but ALL final frames are purely CPU-rendered because GPU’s don’t have enough RAM to render final quality frames) and have consistently found that CPU’s outperform GPU’s for certain computational tasks.
The other issue with GPU’s is that the have severely limited memory sizes in comparison to CPU’s.
So I remain to be convinced that GPU audio is a reasonable thing to spend time on for a tool like Vital.
Also, not everyone has a GPU with more power than their CPU. Modern CPU’s are insanely powerful, and often have a considerable number of unused threads, and are more than capable of loads of impressive computation, and, unlike GPU’s, everyone has one.