[balt]

Hi,

having waited for well over two years now for MR to support GPU rendering, I think I've read pretty much every article in this forum regarding GPU rendering yet I still don't quite understand what the difficulty is in implementing this for MR. Can anyone with a background in render engine design explain this to me/us?

3D is my hobby, professionally I often deal with rather complex problems in the science / computing world (I'm an Astrophysicist and develop lots of software for that), so I do have a fairly good grasp of the algorithmic challenges of implementing parallel algorithms, however, since MR already supports multiple processors, the hard bits have already been figured out so it should only be a software engineer's task of interfacing the CUDA libraries and the rest comes home free? So (in urgent voice) why is this not here yet?

Or am I missing a very fundamental issue here?

Cheers

- Balt

[Aniki]

I posted this already, but here you go again

http://www.luxology.com/tv/training/view.aspx?id=536

cheers

Aniki

[balt]

Hi Aniki,

This video has indeed escaped me. Thanks heaps, that was very interesting and conclusive!

I suppose some company has been misleading us a little bit when it comes to the computing power of GPU's. I didn't know they could only multiply and not divide, figuratively speaking, which of course will throw a bit of a spanner into the elaborate wheels of a render engine...!

Cheers

- Balt