Efficient parallelization of a three-dimensional Navier-Stokes solver on MIMD multiprocessors

We present an efficient parallelization strategy for speeding up the comput ation of a high-accuracy 3-dimensional serial Navier-Stokes solver that tre ats turbulent transonic high-Reynolds flows. The code solves the full compr essible Navier-Stokes equations and is applicable to realistic large size a erodynamic configurations and as such requires huge computational resources in terms of computer memory and execution time. The solver can resolve the flow properly on relatively coarse grids. Since the serial code contains a complex infrastructure typical for industrial code (which ensures its flex ibility and applicability to complex configurations), then the parallelizat ion task is not straightforward. We get scalable implementation on massivel y parallel machines by maintaining efficiency at a fixed value by simultane ously increasing the number of processors and the size of the problem. The 3-D Navier-Stokes solver was implemented on three MIMD message-passing multiprocessors (a 64-processors IBM SP2, a 20-processors MOSIX, and a 64-p rocessors Origin 2000). The same code written with PVM and MPI software pac kages was executed on all the above distinct computational platforms. The e xamples in the paper demonstrate that we can achieve efficiency of about 60 % for as many as 64 processors on Origin 2000 on a full-size 3-D aerodynami c problem which is solved on realistic computational grids.