Search found 382 matches

by spectral
Wed Feb 12, 2014 1:30 pm
Forum: General Development
Topic: Hair intersection : approaches ?
Replies: 16
Views: 13394

Re: Hair intersection : approaches ?

Thanks, but I don't care about shading... just intersections for now ;-)
by spectral
Tue Feb 11, 2014 11:52 am
Forum: General Development
Topic: Hair intersection : approaches ?
Replies: 16
Views: 13394

Hair intersection : approaches ?

Hi, I'm currently looking for all the different approaches of "hair intersection". So, if you have papers, documentation, reference or anything else... please drop me a line ;-) Any council, comment, feedback is also welcomed.... BTW: I have to implement it on the GPU (I'm just afraid of analytic co...
by spectral
Mon Feb 10, 2014 10:01 am
Forum: General Development
Topic: Hybrid : CPU & GPU strategies
Replies: 14
Views: 11564

Re: Hybrid : CPU & GPU strategies

However, in practice, hybrid rendering will never provide the kind of speedup you can achieve with a full OpenCL rendered (or CUDA or whatever you are using).
Sure, but it is not the goal... any speedup is still fine ;-)

Thx
by spectral
Mon Feb 10, 2014 8:35 am
Forum: General Development
Topic: Hybrid : CPU & GPU strategies
Replies: 14
Views: 11564

Re: Hybrid : CPU & GPU strategies

Sure, But here I'm just looking to "improve" the speed of an existing CPU renderer (only the intersection test) ... all I need to store is the geometry ! So, I agree with your vision of the GPU renderers but notice that with time GPU have more & more memory.... more & more cores etc etc... so it is ...
by spectral
Wed Feb 05, 2014 11:02 pm
Forum: General Development
Topic: OpenCL - ray-triangle intersection
Replies: 5
Views: 7884

Re: OpenCL - ray-triangle intersection

There are several factors in this problem 1) the accuracy : do you need very accurate test or not ? 2) the memory usage : is it important for you ? 3) the speed : is it so important In this problem you have to find the exact match for you. By example you can use the ward method, it seems to me very ...
by spectral
Fri Jan 17, 2014 3:06 pm
Forum: General Development
Topic: Hybrid : CPU & GPU strategies
Replies: 14
Views: 11564

Re: Hybrid : CPU & GPU strategies

Thanks Dietger,

But I have also different set of pinned memory !

Of course, I use the same "command queue" for thread's memory transfer and 1 kernel execution :-P
by spectral
Fri Jan 17, 2014 1:06 pm
Forum: General Development
Topic: Hybrid : CPU & GPU strategies
Replies: 14
Views: 11564

Re: Hybrid : CPU & GPU strategies

I suppose your GPU is spending too much time waiting for the transfers (and the CPU work). One thing you could do is to make the transfers asynchronous: have a double buffer for the rays (2x 200k), and fill one while the other is processed. In CUDA, the processing will suffer very little from all t...
by spectral
Fri Jan 17, 2014 9:21 am
Forum: General Development
Topic: Hybrid : CPU & GPU strategies
Replies: 14
Views: 11564

Hybrid : CPU & GPU strategies

Hi there, I'm currently playing with a CPU & GPU hybrid rendering approach... the goal is to "plug" the current GPU intersection engine into an existing CPU renderer. So, here are the performance I got : Pure GPU : 180 MRPS Hybrid : 12 MRPS The GPU usage is: Pure GPU : 97% Hybrid : 37% So, the hybri...
by spectral
Sun Dec 29, 2013 8:19 pm
Forum: ompf2
Topic: Merry Christmas Ompf2 Members (2013)
Replies: 1
Views: 5402

Re: Merry Christmas Ompf2 Members (2013)

Thx a lot Joe,

Wish you the best too for 2014