Dear developers, I have two user reports of 4.6 b1 GPU runs in which the simulation box size increases or decreases quickly after the protein moves over a PBC boundary, ultimately crashing the simulation. Has anything like this already been reported or should I file a more detailed bug report? Carsten