Read-Write Lock vs primitive Lock()

[Aaron Brady]

On Dec 29, 4:17 am, k3xji <sum... at gmail.com> wrote:
> On 29 Aralýk, 11:52, "Gabriel Genellina" <gagsl-... at yahoo.com.ar>
> wrote:
>
> > En Mon, 29 Dec 2008 05:56:10 -0200, k3xji <sum... at gmail.com> escribió:
>
snip
> > > class wthread(threading.Thread):
> > >     def run(self):
> > >             try:
> > >                 #GLOBAL_LOCK.acquireWrite()
> > >                 #GLOBAL_LOCK.acquire_write()
> > >                 GLOBAL_LOCK.acquire()
> > >                 for i in range(GLOBAL_LOOP_COUNT):
> > >                     GLOBAL_VAR = 4
> > >             finally:
> > >                 #GLOBAL_LOCK.release_write()
> > >                 GLOBAL_LOCK.release()
>
> > Note that the thread acquires the lock ONCE, repeats several thousand
> > times an assignment to a *local* variable called GLOBAL_VAR (!)
snip
> > class wthread(threading.Thread):
> >       def run(self):
> >           global GLOBAL_VAR
> >           for i in xrange(GLOBAL_LOOP_COUNT):
> >               GLOBAL_LOCK.acquire()
> >               try:
> >                   GLOBAL_VAR += 1
> >               finally:
> >                   GLOBAL_LOCK.release()
>
> With that, primitive locks perform 10 times better than Read-Write
> lock. See above.
snip

Gabriel's point (one of them) was that 'GLOBAL_VAR' isn't global in
your example.  Your 'wthread' objects aren't sharing anything.  He
added the 'global GLOBAL_VAR' statement, which is important