Towards smaller checkpoints for better intermittent computing: poster abstract