Extending the VLIW approach to fine grain parallelism exploitation beyond single control unit machines