Currently floater_convert doesn't have the datatype as input. So It breaks when encountering 64bit data