Up to Specific issues in real coupled models
I just wanted to check if anyone knows of, or even suspects, a maximum practical limit to the number of component models which can be coupled using OASIS3-MCT in a given coupled model configuration. At the Met Office we're fairly routinely using ocean-atmos-rivers-xios and occasionally wave models, but people are starting to look at coupling more and more things as separate executables (e.g. land-surface, ocean biogeochemistry). My understanding has always been that there's no theoretical limit to the number of component models, and that any limitations were more likely to come from the amount of data being coupled and the related MPI operations (and deadlock avoidance!) etc. Any thoughts appreciated. Richard
Hi Richard, I should have answered your private mail faster, sorry about this. But thanks for asking your question though the forum, it will be interesting to know if any user has encountered a problem with respect to the maximum number of models. I do not foreseen any limitations regarding the number of models. In fact, oasis does not even know how many models are involved (the namcouple keyword NBMODEL is deprecated). It just knows how many fields are exchanged in the global communicator. And I do not see any theoretical limit for the number of fields. Or more precisely as the number of fields is stored in an integer ig_nfield (SELECTED_INT_KIND(9)), I suppose the limit is 2,147,483,647 ! Have a good day, Sophie
Hi Sophie, Thanks a lot for that. (Sorry for my duplicated question, I wasn't sure if the oasishelp email had been replaced by the discussion forum). That's pretty much what I was hoping to hear! It would be interesting if anyone has any experiences to add to the discussion. I can see a scenario where we might end up with, say, 8 different components in the not too distant future. Best wishes, Richard
