Abstract
The development of superfast computers, which operate on data other than on the purely sequential principle, calls for new computational algorithms which make full use of the machine capabilities. Using the inviscid compressible supercritical flow over a circular cylinder as an example, a computational algorithm is developed for a parallel processing computer. All boundary conditions are satisfied explicitly and boundary points are computed in parallel with interior points, thereby greatly increasing the “machine utilization.” Treatment of mesh size changes with the parallel algorithm has been accomplished in a novel way. The influence of various finite difference formulations of the wall boundary condition on the computed results is studied and the results compared with those obtained by other authors. The importance of correlating the formulation of the physical problem with the numerical technique and the computational procedure to secure a high machine utilization in parallel computation is demonstrated.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call