[boost] boost::shared_mutex performance

Subject: [boost] boost::shared_mutex performance
From: Tim Blechmann (tim_at_[hidden])
Date: 2009-04-05 15:31:02

hi all,

i did some performance tests of boost::shared_mutex and a wrapper class
to posix pthread_rwlock_t, which provides most of the interface from the
sharedlockable concept (without the timed locking, which should be easy
to add).

i am not sure, whether the boost::shared_mutex class is supposed to give
some guaranties (like fairness), that my wrapper implementation doesn't
provide ... it does outperform the boost::shared_mutex implementation by
a factor of about 2.5, though (linux, x86_64)

benchmark code and wrapper class are attached. seeing these speedups,
would it be reasonable to introduce a boost::native_shared_mutex class,
based on a native implementation?

best, tim

-- 
tim_at_[hidden]
http://tim.klingt.org
The first question I ask myself when something doesn't seem to be
beautiful is why do I think it's not beautiful. And very shortly you
discover that there is no reason.
  John Cage.