do we have any latency issue on the ML?