Title should be: "Benchmarking OS X semaphores vs message passing". Shared memory + spinlocks should be much faster than shared memory + semaphores (much fast

Title should be: "Benchmarking OS X semaphores vs message passing". Shared memory + spinlocks should be much faster than shared memory + semaphores (much faster, if implemented properly, e.g. spinlock + timeout, on timeout queue another message, again spinlock and try to copy all the elements, etc.).