M × N Communication and Parallel Interpolation in Community Climate System Model Version 3 Using the Model Coupling Toolkit

Abstract

The Model Coupling Toolkit (MCT) is a software library for constructing parallel coupled models from individual parallel models. MCT was created to address the challenges of creating a parallel coupler for the Community Climate System Model (CCSM). Each of the submodels that make up CCSM is a separate parallel application with its own domain decomposition, running on its own set of processors. This application contains multiple instances of the M × N problem, the problem of transferring data between two parallel programs running on disjoint sets of processors. CCSM also requires efficient data transfer to facilitate its interpolation algorithms. MCT was created as a generalized solution to handle these and other common functions in parallel coupled models. Here we describe MCT’s implementation of the data transfer infrastructure needed for a parallel coupled model. The performance of MCT scales satisfactorily as processors are added to the system. However, the types of decompositions used in the submodels can affect performance. MCT’s infrastructure provides a flexible and high-performing set of tools for enabling interoperability between parallel applications.

Ocean–atmosphere–wave characterisation of a wind jet (Ebro shelf, NW Mediterranean Sea)

“…The SWAN model solves the wave action balance equation simulating wind generation and propagation in deep and coastal waters. The modelling system COAWST includes the coupler Model Coupling Toolkit (MCT; Jacob et al, 2005) for the transmission and transformation of the physical variables using a parallel computing approach. The COAWST system also allows for the exchange of data fields on different grids using the Spherical Remapping Interpolation Package (SCRIP; Jones, 1999) to compute the interpolation weights.…”

PyCCSM: Prototyping a python-based community climate system model

“…We believe that once pycpl is running reliably on multiple processors, we can exploit its scalability to reduce the current Python overhead to an acceptable level. We base this prediction on mct's previously demonstrated scalability [8,10] and on the low measured overheads due to Babel on mct's multilingual interfaces [12]. This, combined with the fact the cpl6 coupler is idle much of the time, awaiting data from the ccsm client models, can be used to achieve acceptable performance.…”

“…Numerous Python interfaces to mpi exist, including pympi, 7 pypar, 8 Scientific Python, 9 and mympi. 10 For ccsm, we must support a multiexecutable application that bridges both Fortran and Python.…”

“…Virtually all of the code base described thus far is implemented in Fortran90, and much of it uses Fortran90 derived types and pointers. Greater detail on mph, mct, and cpl6 was provided by He and Ding [7], Larson and Jacob [10,8], and Craig et al [2], respectively. Figure 2 summarises the class hierarchy and functionalities of the mct and cpl6 classes used in ccsm.…”

“…In our past work, we developed significant parallel coupling infrastructure, including the application independent coupling middelware package the Model Coupling Toolkit (mct 2 ) [8,10] and the Community Climate System Model's 3 coupler cpl6 [2]. cpl6 is built on the cpl6 toolkit [2], which offers application specific classes and methods that ease significantly the construction of flux couplers.…”