Dear OpenMPI list,

I am using  a MPI-parallelized simulation program,  with a domain-decomposition in 6-Dimensions.
In order to improve the scalability of my program I would like to know according to what preferences
is MPI distributing the ranks when using MPI_Cart_create( reorder allowed).

To explain my inquiry, imagine a 3-dimensional solver in  X-Y-M and 4 computing
nodes, each nodes consist of 4 Quad-Core CPUs (4(Node)x[ 4(CPUs) x 4(Cores))] CPUs=64CPUs). 

Now I decompose  all 3 dimensions by 4 (4x4x4 = 64) using  MPI_Cart_create. 
MPI has now several  possibilities to map the problem e.g. X-M (locally) on a node and
Y across the nodes, or Y-M (locally) and X across the nodes. 

Now my question is, how can I tell MPI that I want to distribute X-Y locally while 
M is distributed across nodes. The reason is that X-Y 
communication ratio is much large (FFT) compared to M where we have only 
2 communications per time-step via an Allreduce. 
An MPI implementation for the BlueGENE for example has an option 
called mapfile where on can tell MPI how to map the dimensions onto 
the Nodes. I did not found somethings similar for openmpi. 

Does anybody know how to achieve this mapping or could anybody 
tell me where I could find some examples or tutorials ? 

Thank you very much for your help and best wishes 

Paul Hilscher