mem-reordering
An example showing how processors can re-order memory accesses. Much of the code is taken from: http://preshing.com/20120515/memory-reordering-caught-in-the-act/
This is an implementation of the example in section 8.2.3.4 of the Intel 64 and IA-32 Architectures Software Developer’s Manual Volume 3: System Programming Guide. It detects when memory accesses have been reordered, and also shows how the issue goes away with a memory fence. On Linux, it will also build a third binary that ties the threads to a single CPU so that you can see that the problem does not exist when the threads are scheduled onto the same cpu.
Note: on Mac OS X, we don't have fine-grained control over where the threads get scheduled. Mac's idea of affinity is actually L2 cache affinity, meaning that the threads would get scheduled to run out of the same L2 cache, but could run on different cores within the same physical processor. In that case, the re-ordering can still happen.