Page 1 of 1

Question about architecture

Posted: Mon Mar 22, 2010 11:11 pm
by TommyPathTracer
Hi Guy's

I am a bit new at programming on CUDA. I am working on a hobby project, a path tracer. I was wondering how you guy's deal with CUDA.

Do you guys perform all graphics related computing on the GPU or is it more of a hybrid approach? For instance performing ray intersection test on the GPU and doing the BSDFs on the CPU (host)?

I hope you can answer my questions since I am curious how applications like yours and for instance Arion deal with this.

Thanks in advance,

Thomas Kroes

Re: Question about architecture

Posted: Mon Mar 22, 2010 11:26 pm
by James
As this is commercial software I doubt you are going to get much information on how it works/is made.

Re: Question about architecture

Posted: Mon Mar 22, 2010 11:36 pm
by radiance
James wrote:As this is commercial software I doubt you are going to get much information on how it works/is made.
Yeah, sorry, it's our policy not to talk about internals.
What i can say is that octane's raytracer runs entirely on the GPU.
The CPU only does IO and user interface.

Radiance

Re: Question about architecture

Posted: Tue Mar 23, 2010 5:13 am
by YoonKyung
radiance wrote:
James wrote:As this is commercial software I doubt you are going to get much information on how it works/is made.
Yeah, sorry, it's our policy not to talk about internals.
What i can say is that octane's raytracer runs entirely on the GPU.
The CPU only does IO and user interface.

Radiance
I think, this is the winning point of Octane. :D