Parallel performance of a lattice-Boltzmann/finite element cellular blood flow solver on the IBM Blue Gene/P architecture Original Research Article

We discuss the parallel implementation and scaling results of a hybrid lattice-Boltzmann/finite element code for suspension flow simulations. This code allows the direct numerical simulation of cellular blood flow, fully resolving the two-phase nature of blood and the deformation of the suspended phase. A brief introduction to the numerical methods employed is given followed by an outline of the code structure. Scaling results obtained on Argonne National Laboratories IBM Blue Gene/P (BG/P) are presented. Details include performance characteristics on 512 to 65,536 processor cores.