Would there be any reason why the closest hit program is called when a RTgeometrygroup object is passed to the ray generation function while it wouldn’t when a RTgroup object with the same RTgeometrygroup as its only child is used?
Should both work as top level RTobject in the ray.
What’s your Acceleration structure builder on the Group and GeometryGroup?
If they are different, does it work when using Trbvh for both?
If there is nothing else in the scene which requires the Group, leave it away for better performance.
That’s what I guessed, I am using Trbvh for both, using different acceleration types for each doesn’t seem to make a difference.
For the moment the top-level group is not strictly required so I’ll leave it out for now and get back to you if I find anything else.
If you could isolate this in a minimal reproducer program, we could take a look at failing the project sources. Another way to provide a reproducer would be to trace an OptiX API Capture (OAC).
Explanation how to generate one is in the last post here: https://devtalk.nvidia.com/default/topic/803116/error-with-rtpmodelupdate/
Please provide the usual system configuration information when reporting issues:
OS version, installed GPU(s), amount of VRAM on the graphics board(s), display driver version, OptiX version (major. minor.micro), CUDA toolkit version used to generate the PTX input sources, host compiler version, etc.