How openmosix works is that it does load balancing.
Say if you have 2 machines in a OpenMosix cluster and on one machine you have 2 CPU intensive long-lasting programs running and on the other machine you have nothing running.
What Omosix will do is automaticly migrate one of those proccesses so that both machines are running one.
It won't do anything like turn a bunch of machines into a big SMP machine or anything like that. All it does is process migration. It can't even do threads... if a application is multithreaded it can only use one machine. It can only migrate 'forks', which is what you have in Linux when you start a new application.
No single process will ever run faster then the fastest machine in your cluster.
So say your doing a 3D animation rendering.. if your using a rendering method were you go start to finish in one big go then it will only run on one machine. But if you divide up the rendering... say into 3 minutes chunks... with each running as a seperate process then you can start all of them at the same time and OpenMosix can migrate them to other machines.
If you want to try it with Windows applications you might be able to start up a bunch of different instances of your program under Wine.... But Openmosix won't take a bunch of machines and turn them into a single super-machine.
Most High Performance Computing clustering in Linux uses custom applications programmed specificly for that cluster. They use special message-passing libraries and shared storage and batch control programs to take care of automating the jobs. Generally speaking normal applications (say mpeg4 encoding) will have very little benifit from clustering unless the program your using is specially programmed to take advantage of it. With higher-end video and 3d rendering you will have applications that are specially designed to take advantage of a cluster. In these sort of cases OpenMosix isn't going to help you any.