void ThreadBase::notify_one() {
- std::unique_lock<std::mutex>(this->mutex);
+ std::unique_lock<Mutex>(this->mutex);
sleepCondition.notify_one();
}
void ThreadBase::wait_for(volatile const bool& condition) {
- std::unique_lock<std::mutex> lk(mutex);
+ std::unique_lock<Mutex> lk(mutex);
sleepCondition.wait(lk, [&]{ return condition; });
}
// Pick and init the next available split point
SplitPoint& sp = splitPoints[splitPointsSize];
+ sp.spinlock.acquire(); // No contention here until we don't increment splitPointsSize
+
sp.master = this;
sp.parentSplitPoint = activeSplitPoint;
sp.slavesMask = 0, sp.slavesMask.set(idx);
sp.nodes = 0;
sp.cutoff = false;
sp.ss = ss;
-
- // Try to allocate available threads and ask them to start searching setting
- // 'searching' flag. This must be done under lock protection to avoid concurrent
- // allocation of the same slave by another master.
- Threads.spinlock.acquire();
- sp.spinlock.acquire();
-
sp.allSlavesSearching = true; // Must be set under lock protection
+
++splitPointsSize;
activeSplitPoint = &sp;
activePosition = nullptr;
+ // Try to allocate available threads
Thread* slave;
while ( sp.slavesMask.count() < MAX_SLAVES_PER_SPLITPOINT
- && (slave = Threads.available_slave(activeSplitPoint)) != nullptr)
+ && (slave = Threads.available_slave(&sp)) != nullptr)
{
- sp.slavesMask.set(slave->idx);
- slave->activeSplitPoint = activeSplitPoint;
- slave->searching = true; // Slave leaves idle_loop()
- slave->notify_one(); // Could be sleeping
+ slave->spinlock.acquire();
+
+ if (slave->can_join(activeSplitPoint))
+ {
+ activeSplitPoint->slavesMask.set(slave->idx);
+ slave->activeSplitPoint = activeSplitPoint;
+ slave->searching = true;
+ }
+
+ slave->spinlock.release();
}
// Everything is set up. The master thread enters the idle loop, from which
// The thread will return from the idle loop when all slaves have finished
// their work at this split point.
sp.spinlock.release();
- Threads.spinlock.release();
Thread::idle_loop(); // Force a call to base class idle_loop()
assert(!searching);
assert(!activePosition);
+ searching = true;
+
// We have returned from the idle loop, which means that all threads are
- // finished. Note that setting 'searching' and decreasing splitPointsSize must
- // be done under lock protection to avoid a race with Thread::available_to().
- Threads.spinlock.acquire();
+ // finished. Note that decreasing splitPointsSize must be done under lock
+ // protection to avoid a race with Thread::can_join().
sp.spinlock.acquire();
- searching = true;
--splitPointsSize;
activeSplitPoint = sp.parentSplitPoint;
activePosition = &pos;
*bestValue = sp.bestValue;
sp.spinlock.release();
- Threads.spinlock.release();
}
while (!exit)
{
- std::unique_lock<std::mutex> lk(mutex);
+ std::unique_lock<Mutex> lk(mutex);
if (!exit)
sleepCondition.wait_for(lk, std::chrono::milliseconds(run ? Resolution : INT_MAX));
while (!exit)
{
- std::unique_lock<std::mutex> lk(mutex);
+ std::unique_lock<Mutex> lk(mutex);
thinking = false;
void ThreadPool::wait_for_think_finished() {
- std::unique_lock<std::mutex> lk(main()->mutex);
+ std::unique_lock<Mutex> lk(main()->mutex);
sleepCondition.wait(lk, [&]{ return !main()->thinking; });
}
RootMoves.push_back(RootMove(m));
main()->thinking = true;
- main()->notify_one(); // Starts main thread
+ main()->notify_one(); // Wake up main thread: 'thinking' must be already set
}