Hi Sorel,
If I understand you right, you want to access an undefined number of linear
arrays (OpenGL vertex buffers) during a single CUDA-kernel.
Each thread should work exclusively on one of these arrays.
I never tried this myself. Unfortunatelly I think the answer is no.
Here is what I found:
The only way you could call a kernel with a varying number of pointers,
would be to hand over an array of pointers:
Code:
void myKernel( float4** myArrays, uint numArrays )
{
...
}
However, then you will run into the problem that you cannot setup this array of
pointers on the host side as reported here:
http://forums.nvidia.com/index.php?showtopic=87337
Hopefully, I this answer will help you.
Best regards,
Jens
------------------
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=46677#46677
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
