Junction

Discretization of Junction of two pipes (Fig. 3) is presented as the last example. It is a complex example consisting of 120 vertices, 328 curves, 253 surfaces and 58 regions. Again, two truly 3D meshes of different densities have been generated. The coarser one containing 61947 nodes and 54932 elements has been generated using from 1 to 6 processors on SP2 and from 3 to 12 processors on Paramid. The finer mesh consists of 92707 nodes and 110986 elements and has been obtained using from 1 to 8 processors on SP2 and from 6 to 15 processors on Paramid. Although this object is very suitable for parallel discretization using even higher number of processors it was impossible to do that because of temporary lack of resources. The speedup obtained on SP2 machine is drawn in Figure 6 and the time, speedup and efficiency are summarized in Tables 5 and 6. The results achieved on Paramid machine are presented in Figure 9 and in Tables 11 and 12. Again, accurate quantitative comparison of the results is not possible because of indirectly calculated speedups on Paramid. However, also in this example the speedups achieved on SP2 and Paramid exhibit similar profile. A distinct difference may be again observed in the phase of saturation of the speedup which occurs significantly earlier in the case of SP2.