Concurrency support in C++11
Before C++11, the C++ standard made no mention of concurrency. Of course, in practice, programmers wrote multi-threaded and distributed programs in C++ long before 2011. What made that possible was the fact that the compiler writers have voluntarily adopted additional restrictions and guarantees, usually by way of complying with both the C++ standard (for the language) and another standard, such as POSIX, for concurrency support.
C++11 has changed that by introducing the C++ memory model. The memory model describes how threads interact through memory. For the first time, the C++ language was on a solid foundation about concurrency. The immediate practical impact, however, was rather muted since the new C++ memory model was quite similar to the memory models already supported by most compiler writers. There were some subtle differences between those models, and the new standard finally guaranteed the portable behavior of the programs that encounter these...
