I would like to be able to run an application entirely in native mode across multiple nodes and MIC cards. Is it possible?
Yes. You need to build your MPI program in native mode, download it to all the Intel(R) Xeon Phi(TM) coprocessors you want to test. And run the "mpiexec.hydra" command from your host. Please refer to the document http://software.intel.com/en-us/articles/using-the-intel-mpi-library-on-intel-xeon-phi-coprocessor-systems