We present a method of parallelization for computationally extensive problems on multiprocessor machines using a multitasking operating system and a regular compiler. The idea is to divide a process into independent tasks that represent as many execution threads distributed among the available processors. For the distribution of threads on different processors we offer two solutions one in which the operating system is doing this task and one in which it is the programmer’s job to specify which thread runs on which processor. As a model we choose the unsteady flow of an incompressible, viscous fluid through a channel with an obstacle inside. For an easy understanding of the method and for the purpose of outlining the general principles of such flows we considered the simplest geometry: a straight tube with an obstacle inside parallel to its walls.

This content is only available via PDF.
You do not currently have access to this content.