Hi,
On 23/10/2011 04:25, brian tse wrote:
Hi,
I am very new to CUDA and OSG, please accept my apologies if this is
a stupid questions. I am trying to build SPH algorithm using Cuda to
calculate the particle forces and update its position. I would like
to ask is there a way in Cuda to pass the position array to OSG
without doing a cudaMemcpyDeviceToHost operation on every loop?
Is there some special function in OSG that directly integrated into
CUDA?
I'm quite sure there is no such function in plain OSG. You can have a
look at osgPPU or osgCompute node kits that allows one to pass data
between OSG/CUDA. I've only used the texture passing functions, but you
should look for code that passes VBOs between CUDA/OSG. Consult the CUDA
docs on the sections about OpenGL interop.
I have a CPU loop for CUDA kernel, and I have another CPU loop for
OSG. When the OSG need the graphics update, it send a flag to CUDA
reqesting for new position array. Is this the right way to do it?
You should be able to avoid copy to host. Let CUDA update the VBO that
OSG then uses. The synchronisation might be an issue, that's why it
might be better to use cuda inside the scene graph.
cheers
jp
Thank you very much in advance. I really appreciate your effort.
Cheers, Brian
-- Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=43510#43510
___ osg-users mailing
list osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
--
This message is subject to the CSIR's copyright terms and conditions, e-mail legal notice, and implemented Open Document Format (ODF) standard.
The full disclaimer details can be found at http://www.csir.co.za/disclaimer.html.
This message has been scanned for viruses and dangerous content by MailScanner,
and is believed to be clean.
___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org