How to use onnxruntime on same GPU as graphics api i.e opengl

Hi does anyone know what the correct way to architect something like this is? It seems like some gpus can't interrupt the long running onnx operations in order to prepare frame for rendering. Is there a way to tell if a