The memory coherence scenarios we consider always involve multiple processors. If we only have one processor, even if its multi-threaded, there would not be a shared address space abstraction among the threads.