[manuutin]

Greetings.

I've developed advanced user coding in star-cd and so far it has been enough to work with only small cases and serial runs, i.e., with a single core. Now the case is so large I have to go up to parallel runs.

My problem is cell indexing. In my models, I need to record data from a cell (i) and all its neighbors. Now I assume that in each process the cells are numbered from i =1 to i=N/M/L and so on. If I'm in cell (i) in a given process, how do I get the data, e.g. T from the neighbouring cell if it lies in another processor's partition? Is there perhaps a 'global mesh' or some sort of mapping I could utilize?

Any advice is highly appreciated!

BR, Mika