The CCSM system consists of 5 separate executables, one of which is the Coupler. A series of c-shell scripts and GNU Makefiles are used to build CCSM including the Coupler. Usage of the scripts that build CCSM is described in detail in the CCSM 3.0 User's Guide, and the Coupler user should consult this guide for information on building the complete system.
The CCSM Coupler requires two libraries, the Model Coupling Toolkit (MCT) and MPH3. These libraries are distributed with CCSM and are compiled before the Coupler within the CCSM3 build system. These libraries must compile successfully for the Coupler to complete it's build.
The Coupler can only execute within the complete CCSM system. In CCSM3, this requires atmosphere, ice, land, and ocean models along with the Coupler. There is no ``standalone'' Coupler. Again, the CCSM3.0 User's Guide should be consulted for information on how to run CCSM3. CCSM3 allows several possible combinations of active and data models to be executed as a coupled system with the Coupler. The simplest system to run the Coupler with is to use all data models for the ocean, atmosphere, land and sea ice. This will still result in a 5 executable system but the data models are very simple and each run on only 1 processor. For advanced testing, an all-dead models case can be run with the Coupler. Unlike the data models, the dead models do not produce physically realistic fields but they can be run on any number of processors. An all-dead-model case is useful for testing the scalability of the model-coupler communication routines.
The Coupler can run on any number of processors. After creating a case (See CCSM3 Users Guide), edit the env_mach.$MACH file and change ntasks_cpl to change the number of MPI processors the Coupler runs on. nthrds_cpl should always be left as 1 because the Coupler currently has no thread-level parallelism.