On the parallelization of a global climate-chemistry modeling system

Coupled climate-chemistry simulations are computationally intensive owing t o the spatial and temporal scope of the problem. In global chemistry models , the time integrations encountered in the chemistry and aerosol modules us ually comprise the major CPU consumption. Parallelization of these segments of the code can contribute to multifold CPU speed-ups with minimal modific ation of the original serial code. This technical note presents a single pr ogram-multiple data (SPMD) strategy applied to the time-split chemistry mod ules of a coupled climate - global tropospheric chemistry model. Latitudina l domain decomposition is adopted along with a dynamic load-balancing techn ique that uses the previous time-step's load/latitude estimates for distrib uting the latitude bands amongst the processors. The coupled model is manua lly parallelized using the Message Passing Interface standard (MPI) on a di stributed memory platform (IBM-SP2), Load-balancing efficiencies and the as sociated MPI overheads are discussed. Overall speed-ups and efficiencies ar e also calculated for a series of runs employing up to eight processors. (C ) 1999 Elsevier Science Ltd. All rights reserved.