Hello, I'm trying to use osgCompute in a manner very similar to the osgRTTDemo. The only difference is that my source texture is a float32 and the target texture is an unsigned 8 bit int texture. I have it set up so that if I keep the power of 16 texture size it works perfectly. However, if I change the texture size and make what I think are the appropriate changes to the TexFilter.cu code, I get black spots in my output texture that make it look like the correct data is being written to the incorrect output texture location in memory. The changes I made were to round up the number of blocks and to put an if statement to only process threads with x and y values within the image bounds, which from my experience looking at CUDA code is a very common practice.
So, the fundamental questions is: how can the power of 16 texture limitation be removed from the osgRTTDemo ? ------------------ Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=52241#52241 _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
