Path: csiph.com!x330-a1.tempe.blueboxinc.net!newsfeed.hal-mli.net!feeder1.hal-mli.net!weretis.net!feeder1.news.weretis.net!news.albasani.net!.POSTED!not-for-mail
From: BGB <cr88192@hotmail.com>
Newsgroups: comp.lang.c,comp.programming.threads,comp.unix.programmer
Subject: Re: POSIX threads hard to implement in Windows
Date: Wed, 03 Aug 2011 07:32:11 -0700
Organization: albasani.net
Lines: 42
Message-ID: <j1bme3$vsa$1@news.albasani.net>
References: <iuso52$shb$1@dont-email.me> <iussos$ndi$1@speranza.aioe.org> <97ek3iFq2bU2@mid.individual.net> <4e13c801$0$20925$a729d347@news.telepac.pt> <lnwrfw1679.fsf@nuthaus.mib.org> <4e13d4a7$0$20921$a729d347@news.telepac.pt> <iv0ljc$fce$1@dont-email.me> <jobee8-aki.ln1@jones.homeip.net> <lnbox70xby.fsf@nuthaus.mib.org> <37pfe8-4gl.ln1@jones.homeip.net> <iv42dd$ed2$2@dont-email.me> <iv4blu$79l$1@dont-email.me> <c1ahe8-hi.ln1@wilbur.25thandClement.com> <97mi1gFbcvU4@mid.individual.net> <487c17t7h0f64f1mjmm83gj5f080o56cfa@4ax.com> <iv5p2f$95u$1@dont-email.me>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Trace: news.albasani.net lJ500u9jVaTRLMgIHkwuTOJ6cP0rouOez8lMsOGJG1iW5K+X2EQ3uvaUGXEzLG2xbN+bQ3/uJ+Dz8JtRuhaQoA==
NNTP-Posting-Date: Wed, 3 Aug 2011 14:36:51 +0000 (UTC)
Injection-Info: news.albasani.net; logging-data="ju+vP/U8uXH2aWF83QXYeMhqnkX5zoY4020A2Rxs3UwWkfkmadjteryJ6qGv15khkYftjgMACpFBW7usQgsX27Y1HNRi4uWLrh77BxwzwfOmQyBFAJaaJ8QCrr3Xs3I5"; mail-complaints-to="abuse@albasani.net"
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:5.0) Gecko/20110624 Thunderbird/5.0
In-Reply-To: <iv5p2f$95u$1@dont-email.me>
Cancel-Lock: sha1:BbpueZItqk63sh2d0XLKD64A5QE=
Xref: x330-a1.tempe.blueboxinc.net comp.lang.c:9006 comp.programming.threads:485 comp.unix.programmer:1053

On 7/7/2011 7:12 PM, MikeP wrote:
> Robert Wessel wrote:
>> On Fri, 08 Jul 2011 08:07:30 +1200, Ian Collins<ian-news@hotmail.com>
>> wrote:
>
>>> Does the windows API support process shared mutexes?  The C1x
>>> proposal doesn't include them probably because the process model is
>>> beyond the C standard.
>>
>>
>> Sure.  Most of the synchronization objects in Windows are system wide.
>
> CAN be. If you name them, THEN they can be used amongst processes.
> Unnamed synch objects are intra-process. Whether having unnamed synch
> objects makes them lighter-weight (higher performance) in usage, I don't
> know. Simple enough to profile though.
>

these mutexes are very expensive, almost absurdly slow. a lock/unlock 
cycle takes around 1-2us, which is far longer than most operations one 
may reasonably want to lock (on a reasonably modern computer, this is 
around 3000 to 6000 clock cycles).

"Critical Sections" are a bit faster, and so are preferable to the 
generic mutex objects (but, much faster than dead slow is still not 
necessarily fast).


much faster IME was to implement a spinlock-with-sleeep operation (using 
"InterlockedExchange", ...), which can generally handle 
locking/unlocking in the low ns range (more around 5ns or so).

the above fast mutexes make a lot more sense when ones' goal is 
something like:
lock mutex;
add entry to cross-thread linked list (such as a message queue);
unlock mutex.

or: the corresponding action to remove something from a linked-list.


rather than... whatever the hell MS intends their mutexes for...