max gpu's possible?

[g0ll4m]

What is the number (if there is one) of GPU's that Octane can utilize now that there's network capabilities.

Can I theoretically have 10 networked machines that each have 4 GTX 780's? And Octane will use all CUDA power?

So roughly 100,000 CUDA cores.

[glimpse]

12 GPUs is the limit for now, so in theory You can have a beast with dual GPUs on a server grade mobo and basicly max out this number..or..build 3x 4GPU/system (lets say with quad Titans or 780s) - that way You need three licences & could do network rendering with those.

They are working on Licence side, so with time maybe we'll see different option to be presented that will enable to utilise more GPUs via netorked machines (but for now 12 GPUs is the limmit).

[Octopot]

Would each render slave be able to utilize all available vram or does it still need to have 1 gpu card load data?

[glimpse]

scene should fit into every card..- so no matter how much You have You will always be limited by the active card having smallest amount of vRAM in Your system - as that's going to be the ceiling You're going to see in the program =)