Dear DFTB+ users, Do someone have experience or comments on massively parallel DFTB+ calculations? Is the gain interesting with hundred atoms systems (eventually with some k points) on hundreds of cores? Best regards S. Nénon