The article presents an algorithmic model of sound propagation in rooms to run on parallel and distributed computer systems. This algorithm is used by the authors in an implementation of an adaptable high-performance computer system simulating various fields and providing scalability on an arbitrary number of parallel central and graphical processors as well as distributed computer clusters. Many general-purpose computer simulation systems have limited usability when it comes to high-precision simulation associated with large numbers of elementary computations due to their lack of scalability on various parallel and distributed platforms. The more the required adequacy of the model is, the higher the numbers of steps of the simulation algorithms are. Scalability permits a use hybrid parallel computer systems and improves efficiency of the simulation with respect to adequacy, time consumptions, and total costs of simulation experiments. The report covers such an algorithm which is based on an approximate superposition of acoustical fields and provides adequate results, as long as the used equations of acoustics are linear. The algorithm represents reflecting surfaces as sets of vibrating pistons and uses the Rayleigh integral to calculate their scattering properties. The article also provides a parallel form of the algorithm and analysis of its properties in parallel and sequential forms.