Path: ...!news.misty.com!weretis.net!feeder9.news.weretis.net!news.quux.org!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!eternal-september.org!.POSTED!not-for-mail
From: "Chris M. Thomasson" <chris.m.thomasson.1@gmail.com>
Newsgroups: comp.lang.c++
Subject: Re: signalling a condvar from inside vs. signalling a condvar von
 outside
Date: Thu, 24 Apr 2025 13:05:08 -0700
Organization: A noiseless patient Spider
Lines: 74
Message-ID: <vue5hl$2aggi$1@dont-email.me>
References: <vte0g6$pmgv$1@raubtier-asyl.eternal-september.org>
 <vtpcnt$384n4$1@dont-email.me>
 <vtpu86$3qnsh$1@raubtier-asyl.eternal-september.org>
 <vtq3dt$3vg77$1@dont-email.me>
 <vtqo57$hi7r$1@raubtier-asyl.eternal-september.org>
 <vtrf3u$15mgp$1@dont-email.me>
 <vtsph1$2d55o$1@raubtier-asyl.eternal-september.org>
 <vtu9un$3mhje$1@dont-email.me>
 <vtub8i$3o27s$1@raubtier-asyl.eternal-september.org>
 <nPOMP.335105$j2D.258162@fx09.iad>
 <vu0f8m$1nm7v$2@raubtier-asyl.eternal-september.org>
 <eVVMP.1467426$BrX.929148@fx12.iad> <vu3obr$kp0a$2@dont-email.me>
 <vu4j24$1fq9b$1@raubtier-asyl.eternal-september.org>
 <vu6nrs$3b29s$1@dont-email.me>
 <vu78mg$3t3v4$1@raubtier-asyl.eternal-september.org>
 <vu925c$1esv8$1@dont-email.me>
 <vu9uqk$2bldm$1@raubtier-asyl.eternal-september.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Injection-Date: Thu, 24 Apr 2025 22:05:10 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="9ab23ff6e21fe9bbc243a2a98cfc6588";
	logging-data="2441746"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX18MRInDQXY+tebba7iRIUCdksjd2Bh3V88="
User-Agent: Mozilla Thunderbird
Cancel-Lock: sha1:yTodaUE5Q2/ekfo4+jUnzVp3KOA=
Content-Language: en-US
In-Reply-To: <vu9uqk$2bldm$1@raubtier-asyl.eternal-september.org>
Bytes: 4860

On 4/22/2025 10:46 PM, Bonita Montero wrote:
> Now I wrote a little program to test if there's thundering herd problem
> with glibc's mutex / condvar. This it is:
> 
> #include <iostream>
> #include <thread>
> #include <mutex>
> #include <condition_variable>
> #include <atomic>
> #include <semaphore>
> #include <vector>
> #include <sys/resource.h>
> 
> using namespace std;
> 
> int main()
> {
>      constexpr size_t N = 10'000;
>      int nClients = thread::hardware_concurrency() - 1;
>      mutex mtx;
>      int signalled = 0;
>      condition_variable cv;
>      atomic_int ai( 0 );
>      binary_semaphore bs( false );
>      vector<jthread> clients;
>      atomic_int64_t nVoluntary( 0 );
>      for( int c = nClients; c; --c )
>          clients.emplace_back( [&]
>              {
>                  for( size_t r = N; r; --r )
>                  {
>                      unique_lock lock( mtx );
>                      cv.wait( lock, [&] { return (bool)signalled; } );
>                      --signalled;
>                      lock.unlock();
>                      if( ai.fetch_sub( 1, memory_order_relaxed ) == 1 )
>                          bs.release( 1 );
>                  }
>                  rusage ru;
>                  getrusage( RUSAGE_THREAD, &ru );
>                  nVoluntary.fetch_add( ru.ru_nvcsw, memory_order_relaxed );
>              } );
>      for( size_t r = N; r; --r )
>      {
>          unique_lock lock( mtx );
>          signalled = nClients;
>          cv.notify_all();
>          ai.store( nClients, memory_order_relaxed );
>          lock.unlock();
>          bs.acquire();
>      }
>      clients.resize( 0 );
>      cout << N << " rounds," << endl;
>      cout << (double)nVoluntary.load( memory_order_relaxed ) / nClients 
> << " context switches pe thread" << endl;
> }
> 
> It spawns one less threads than ther are hardware threads. These
> all wait for a condvar and a counter which is initially the number
> of threads and that must be > 0 for the wait to succeed. This counter
> is decremented by each thread. Then the threads decrement an atomic
> and if it becomes zero the last thread raises a semaphore, thereby
> waking up the main thread.
> This are the results for 10'000 rounds on a 32-thread machine:
> 
>      10000 rounds,
>      2777.29 context switches pe thread
> 
> So there are less context-switches than rounds and there's no
> thundering herd with glibc.
> 

Sigh... Here is a challenge for you. Get it working in a race detector, 
say, Relacy?