High–Level Synthesis for High–Speed

The flow of a liquid around an obstacle can be described by a set of partial differential equations, known as the Navier–Stokes equations. It requires very complex calculations to solve these equations for a given problem. Moreover it is very tedious to describe the boundary conditions for non–trivial problems. Instead of using numerical methods to compute the liquid flow one can also describe the local behaviour of the particles on a molecular level. This model is called a lattice gas and most often implemented by a cellular automaton. The description of the particle behaviour is called the local rule of the cellular automaton. A two dimensional model that shows almost all the properties of the Navier–stokes equations is the FHP–gas[1]. Field sizes of 1 million cells and more are common to get precise results. The application of the local rule to all cells of the field is called a generation. Between 1000 and 10000 generations are required in lattice gas models to reach a stable state. Thus the rule must be applied to the cells 10 to 10 times. This takes several hours in a software simulation and thus special hardware has been developed to speed up this simulation.