Art Tevs wrote:
Just grab it and put instead of the kernel used in that example your FFT code and it should work. There is just one drawback, when using Cuda within OpenGL, the performance. The memory space of Cuda and OpenGL is not shared, hence it costs you some time to copy a texture into cuda space and copy it then back. However I am not sure if there were new extensions to speed up the copy process.
I believe there were some extensions added to support this for OpenCL (clCreateFromGLTexture2D, etc). Chances are if these are supported for OpenCL, they're also there for CUDA (you might need CUDA 3.0 for this, though). Also, possibly look at the NV_copy_image extension.
--"J" _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
