There are three separate issues that "atomic" types in C++11 address:
tearing: a read or write involves multiple bus cycles, and a thread switch occurs in the middle of the operation; this can produce incorrect values.
cache coherence: a write from one thread updates its processor's cache, but does not update global memory; a read from a different thread reads global memory, and doesn't see the updated value in the other processor's cache.
compiler optimization: the compiler shuffles the order of reads and writes under the assumption that the values are not accessed from another thread, resulting in chaos.
Using std::atomic ensures that all three of these issues are managed correctly. Not using std::atomic leaves you guessing, with, at best, non-portable code.
