On 3/24/2014 9:30 AM, Thiago Macieira wrote: > Em seg 24 mar 2014, às 16:27:06, André Somers escreveu: >> The point is, that usually you don't _need_ synchronization for this >> purpose, and using synchronization does come at a cost. It is usually >> not critical that the worker thread stops immediately when signalled, >> just that it stops soon. So even if - due to missing synchronization and >> the thread reading the flag at the same time it being changed - the >> worker does an other iteration before stopping, nothing is lost. All >> that happens is that the thread does a little bit of work that was not >> needed. The bonus is that the rest of the time, no synchronization is >> needed at all, which may result in better performance. > Note that using a regular, non-atomic variable could mean that the thread > takes a LONG time to see the change, depending on the architecture and OS > scheduling.
I actually use the "volatile" tag to tell the compiler that it cannot optimize (i.e., make compile-time assumptions about) the variable that it decorates. Does std::atomic have the same meaning for optimizations? I can't get a clear picture of this by a (cursory) Google. If compilers still consider std::atomic in aggressive optimizing, then it might be an either/or depending upon particular need. _______________________________________________ Interest mailing list Interest@qt-project.org http://lists.qt-project.org/mailman/listinfo/interest
