Parallel performance of the interpolation supplemented lattice Boltzmann method

The interpolation supplemented lattice Boltzmann (ISLB) method was proposed by He et. al. (J. Comput. Phys., 129, 357 (1996)) as a modification of the traditional lattice Boltzmann (LB) method for simulating incompressible flows. The traditional LB method has been shown to be a highly parallel method (Computers in Physics, 8, No. 6,705 (1994)). In this paper, the parallel performance of the ISLB method, which has different computational requirements than the traditional LB method, is investigated. The performance of the ISLB method on Cray X1, Cray T3E-900 and SGI Origin 3000 is presented. The noteworthy feature of the present implementation of the ISLB method is that it was able to achieve a sustained speed of 550 Gflops on a 124 processor Cray X1. © Springer-Verlag Berlin Heidelberg 2003.