Hello lovelies!
What I might be asking could be impossible so I'm asking two questions really, firstly whether it is even possible and secondly, if it is, how should I go about it.
I am working on a bit of safety software for air traffic control that involves screen capturing and post processing. At the moment we are having to use high end Nvidia GPUs that are compatible with their capture sdk, it can grab the screen and move the frame buffer to a cuda core for processing and encoding without touching the CPU or system memory. This is really important because we cannot exceed 10% CPU utilization on 1 core, even when grabbing 4 4k screens and we have similarly restrictive requirements for memory usage.
Can I do this with AMD cards? If so, which cards and with what APIs? My google has not been fruitful. AMD cards tend to be cheaper and much more compatible with Linux so I'd love to support them.
Thanks in advance. No problem if it isn't possible as of now!
James
