Comment by delta_p_delta_x

My solution is to only solve problems that are embarrassingly parallel, like graphics (one pixel = one thread) or physics simulations (one object = one thread), and escape the pain of synchronisation.