phabouzit

3 ms·

1. 1.▲
   by phabouzit 10y ago

   1) indeed there are a couple of barriers that could be optimized 2) in addition to what Phil already said (which is that there's no priority inversion for Safari/WebKit to have as all threads taking these locks have the same prior
2. 2.▲
   by phabouzit 10y ago

   I disagree strongly as too strong barriers on non TSO hardware (such as say, armv7 or armv8) will just cause even more of that memory contention for no good reasons. Also on a contended lock, there's a very strong possibility that you
3. 3.▲
   by phabouzit 10y ago

   Which OS X Release? with which benchmark? uncontended pthread_mutex was made significantly faster in OS X El Capitan (almost twice as fast as before), and on my own mac is only twice as slow as OSSpinLock, and OSSpinLock unlock is a store(l