Sergey Sharybin
c1dd74580e
Tweaked scheduling so it survives this situation by scattering "extra" threads uniformly over all the NUMA nodes. There are still tweaks possible to make some specific hardware configurations work better.
148 lines
3.0 KiB
C++
148 lines
3.0 KiB
C++
/*
|
|
* Copyright 2011-2013 Blender Foundation
|
|
*
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
* you may not use this file except in compliance with the License.
|
|
* You may obtain a copy of the License at
|
|
*
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
*
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
* See the License for the specific language governing permissions and
|
|
* limitations under the License.
|
|
*/
|
|
|
|
#ifndef __UTIL_THREAD_H__
|
|
#define __UTIL_THREAD_H__
|
|
|
|
#include <thread>
|
|
#include <mutex>
|
|
#include <condition_variable>
|
|
#include <functional>
|
|
#include <queue>
|
|
|
|
#ifdef _WIN32
|
|
# include "util_windows.h"
|
|
#else
|
|
# include <pthread.h>
|
|
#endif
|
|
|
|
#ifdef __APPLE__
|
|
# include <libkern/OSAtomic.h>
|
|
#endif
|
|
|
|
#include "util/util_function.h"
|
|
|
|
CCL_NAMESPACE_BEGIN
|
|
|
|
typedef std::mutex thread_mutex;
|
|
typedef std::unique_lock<std::mutex> thread_scoped_lock;
|
|
typedef std::condition_variable thread_condition_variable;
|
|
|
|
/* own pthread based implementation, to avoid boost version conflicts with
|
|
* dynamically loaded blender plugins */
|
|
|
|
class thread {
|
|
public:
|
|
/* NOTE: Node index of -1 means that affinity will be inherited from the
|
|
* parent thread and no override on top of that will happen. */
|
|
thread(function<void()> run_cb, int node = -1);
|
|
~thread();
|
|
|
|
static void *run(void *arg);
|
|
bool join();
|
|
|
|
/* For an existing thread descriptor which is NOT running yet, assign node
|
|
* on which it should be running. */
|
|
void schedule_to_node(int node);
|
|
|
|
protected:
|
|
function<void()> run_cb_;
|
|
std::thread thread_;
|
|
bool joined_;
|
|
int node_;
|
|
};
|
|
|
|
/* Own wrapper around pthread's spin lock to make it's use easier. */
|
|
|
|
class thread_spin_lock {
|
|
public:
|
|
#ifdef __APPLE__
|
|
inline thread_spin_lock() {
|
|
spin_ = OS_SPINLOCK_INIT;
|
|
}
|
|
|
|
inline void lock() {
|
|
OSSpinLockLock(&spin_);
|
|
}
|
|
|
|
inline void unlock() {
|
|
OSSpinLockUnlock(&spin_);
|
|
}
|
|
#elif defined(_WIN32)
|
|
inline thread_spin_lock() {
|
|
const DWORD SPIN_COUNT = 50000;
|
|
InitializeCriticalSectionAndSpinCount(&cs_, SPIN_COUNT);
|
|
}
|
|
|
|
inline ~thread_spin_lock() {
|
|
DeleteCriticalSection(&cs_);
|
|
}
|
|
|
|
inline void lock() {
|
|
EnterCriticalSection(&cs_);
|
|
}
|
|
|
|
inline void unlock() {
|
|
LeaveCriticalSection(&cs_);
|
|
}
|
|
#else
|
|
inline thread_spin_lock() {
|
|
pthread_spin_init(&spin_, 0);
|
|
}
|
|
|
|
inline ~thread_spin_lock() {
|
|
pthread_spin_destroy(&spin_);
|
|
}
|
|
|
|
inline void lock() {
|
|
pthread_spin_lock(&spin_);
|
|
}
|
|
|
|
inline void unlock() {
|
|
pthread_spin_unlock(&spin_);
|
|
}
|
|
#endif
|
|
protected:
|
|
#ifdef __APPLE__
|
|
OSSpinLock spin_;
|
|
#elif defined(_WIN32)
|
|
CRITICAL_SECTION cs_;
|
|
#else
|
|
pthread_spinlock_t spin_;
|
|
#endif
|
|
};
|
|
|
|
class thread_scoped_spin_lock {
|
|
public:
|
|
explicit thread_scoped_spin_lock(thread_spin_lock& lock)
|
|
: lock_(lock) {
|
|
lock_.lock();
|
|
}
|
|
|
|
~thread_scoped_spin_lock() {
|
|
lock_.unlock();
|
|
}
|
|
|
|
/* TODO(sergey): Implement manual control over lock/unlock. */
|
|
|
|
protected:
|
|
thread_spin_lock& lock_;
|
|
};
|
|
|
|
CCL_NAMESPACE_END
|
|
|
|
#endif /* __UTIL_THREAD_H__ */
|