Load-balancing for parallel computation of fluid dynamics problems Original Research Article

Recent developments in computer networks and massively parallel computers provide exciting opportunities for computational fluid dynamics (CFD). In order to reduce the computation time and increase the efficiency of parallel CFD, we developed a new method to balance the load of computers. In this method, we realistically assumed that (1) computers are networked and may have different computation speeds, (2) CFD data can be divided into many small data blocks with possibly different sizes, and (3) the number of computers is much less than the number of data blocks. The greedy method for optimization is employed to provide a near optimal distribution of the data blocks on computers for achieving computer load balancing. This method provides a near optimal time solution to the computation of given parallel CFD algorithms.