These questions may seem very fundamental to many of you. If you ask yourself "Can he really be asking such basic questions?", the answer is yes!
I recently assembled a Linux desktop using an A10-7800 and a GeForce GT 740 card. The whole point of this machine is to give me a testbed for learning more about GPGPU programming. My starting point in this effort was ground-zero... I have no experience with installing video-card drivers. At work, I write OpenCL code for a system that uses multiple Tesla cards but others were responsible for configuring that box so I don't know what magic was involved in setting it up.
After some trial and error, I got it working with the NVidia card driving the display. I am able to write OpenCL and Cuda code that execute on the NVidia card with no problem. Now I would like to enable OpenCL on the GPU cores built-in to the A10. My google searches haven't really pointed to a direct "Here's how you do it" answer though at least one site seemed to say it was possible. The one article I saw that seemed kind of close was geared toward using the APU for driving the display. I don't think that is what I want to do but feel free to poke holes in that opinion.
With all the background out of the way, here are my questions:
1) First, given that my primary goal is to write OpenCL code, would I be better off using the A10 to drive the display or to leave it as-is, with the 740 driving it?
2) Since the X-Server is using the 740, does that imply that some of the memory on that card is unavailable for my OpenCL code? (Yeah, this is a pretty basic question. I know.)
3) Very basic terminology question - I'm ignorant about what Catalyst is used for. Is that the display driver? Is it needed if I simply want to write OpenCL code that runs on the APU?
4) Is it even possible to configure this system in a way that it looks like two GPGPUs are present at the same time?
Hopefully this all makes sense. I appreciate any help you can provide.