Register allocation with instruction scheduling for VLIW-architectures