1、GIL是什么?
GIL的全称是Global Interpreter Lock(全局解释器锁),来源是python设计之初的考虑,为了数据安全所做的决定。
2、每个CPU在同一时间只能执行一个线程(在单核CPU下的多线程其实都只是并发,不是并行,并发和并行从宏观上来讲都是同时处理多路请求的概念。但并发和并行又有区别,并行是指两个或者多个事件在同一时刻发生;而并发是指两个或多个事件在同一时间间隔内发生。)
在Python多线程下,每个线程的执行方式:
获取GIL
执行代码直到sleep或者是python虚拟机将其挂起。
释放GIL
可见,某个线程想要执行,必须先拿到GIL,我们可以把GIL看作是“通行证”,并且在一个python进程中,GIL只有一个。拿不到通行证的线程,就不允许进入CPU执行。
在Python2.x里,GIL的释放逻辑是当前线程遇见IO操作或者ticks计数达到100(ticks可以看作是Python自身的一个计数器,专门做用于GIL,每次释放后归零,这个计数可以通过 sys.setcheckinterval 来调整),进行释放。
而每次释放GIL锁,线程进行锁竞争、切换线程,会消耗资源。并且由于GIL锁存在,python里一个进程永远只能同时执行一个线程(拿到GIL的线程才能执行),这就是为什么在多核CPU上,python的多线程效率并不高。
那么是不是python的多线程就完全没用了呢?
在这里我们进行分类讨论:
1、CPU密集型代码(各种循环处理、计数等等),在这种情况下,由于计算工作多,ticks计数很快就会达到阈值,然后触发GIL的释放与再竞争(多个线程来回切换当然是需要消耗资源的),所以python下的多线程对CPU密集型代码并不友好。
2、IO密集型代码(文件处理、网络爬虫等),多线程能够有效提升效率(单线程下有IO操作会进行IO等待,造成不必要的时间浪费,而开启多线程能在线程A等待时,自动切换到线程B,可以不浪费CPU的资源,从而能提升程序执行效率)。所以python的多线程对IO密集型代码比较友好。
而在python3.x中,GIL不使用ticks计数,改为使用计时器(执行时间达到阈值后,当前线程释放GIL),这样对CPU密集型程序更加友好,但依然没有解决GIL导致的同一时间只能执行一个线程的问题,所以效率依然不尽如人意。
多核性能
多核多线程比单核多线程更差,原因是单核下多线程,每次释放GIL,唤醒的那个线程都能获取到GIL锁,所以能够无缝执行,但多核下,CPU0释放GIL后,其他CPU上的线程都会进行竞争,但GIL可能会马上又被CPU0拿到,导致其他几个CPU上被唤醒后的线程会醒着等待到切换时间后又进入待调度状态,这样会造成线程颠簸(thrashing),导致效率更低
多进程为什么不会这样?
每个进程有各自独立的GIL,互不干扰,这样就可以真正意义上的并行执行,所以在python中,多进程的执行效率优于多线程(仅仅针对多核CPU而言)。
所以在这里说结论:多核下,想做并行提升效率,比较通用的方法是使用多进程,能够有效提高执行效率。
所以,如果不想浪费时间,可以直接看多进程。
直接利用函数创建多线程
Python中使用线程有两种方式:函数或者用类来包装线程对象。
函数式:调用thread模块中的start_new_thread()函数来产生新线程。语法如下:
<span class="s1">thread</span><span class="s2">.</span><span class="s1">start_new_thread </span><span class="s2">(</span> <span class="s3">function</span><span class="s2">,</span><span class="s1"> args</span><span class="s2">[,</span><span class="s1"> kwargs</span><span class="s2">]</span> <span class="s2">)</span>
1
<span class="s1">thread</span><span class="s2">.</span><span class="s1">start_new_thread </span><span class="s2">(</span> <span class="s3">function</span><span class="s2">,</span><span class="s1"> args</span><span class="s2">[,</span><span class="s1"> kwargs</span><span class="s2">]</span> <span class="s2">)</span>
参数说明:
function – 线程函数。
args – 传递给线程函数的参数,他必须是个tuple类型。
kwargs – 可选参数。
先用一个实例感受一下:
-- coding: UTF-8 --
import thread
import time
为线程定义一个函数
def print_time(threadName, delay):
count = 0
while count < 5:
time.sleep(delay)
count += 1
print "%s: %s" % (threadName, time.ctime(time.time()))
创建两个线程
try:
thread.start_new_thread(print_time, ("Thread-1", 2,))
thread.start_new_thread(print_time, ("Thread-2", 4,))
except:
print "Error: unable to start thread"
while 1:
pass
print "Main Finished"
-- coding: UTF-8 --
import thread
import time
为线程定义一个函数
def print_time(threadName, delay):
count = 0
while count < 5:
time.sleep(delay)
count += 1
print "%s: %s" % (threadName, time.ctime(time.time()))
创建两个线程
try:
thread.start_new_thread(print_time, ("Thread-1", 2,))
thread.start_new_thread(print_time, ("Thread-2", 4,))
except:
print "Error: unable to start thread"
while 1:
pass
print "Main Finished"
运行结果如下:
Thread-1: Thu Nov 3 16:43:01 2016
Thread-2: Thu Nov 3 16:43:03 2016
Thread-1: Thu Nov 3 16:43:03 2016
Thread-1: Thu Nov 3 16:43:05 2016
Thread-2: Thu Nov 3 16:43:07 2016
Thread-1: Thu Nov 3 16:43:07 2016
Thread-1: Thu Nov 3 16:43:09 2016
Thread-2: Thu Nov 3 16:43:11 2016
Thread-2: Thu Nov 3 16:43:15 2016
Thread-2: Thu Nov 3 16:43:19 2016
Thread-1: Thu Nov 3 16:43:01 2016
Thread-2: Thu Nov 3 16:43:03 2016
Thread-1: Thu Nov 3 16:43:03 2016
Thread-1: Thu Nov 3 16:43:05 2016
Thread-2: Thu Nov 3 16:43:07 2016
Thread-1: Thu Nov 3 16:43:07 2016
Thread-1: Thu Nov 3 16:43:09 2016
Thread-2: Thu Nov 3 16:43:11 2016
Thread-2: Thu Nov 3 16:43:15 2016
Thread-2: Thu Nov 3 16:43:19 2016
可以发现,两个线程都在执行,睡眠2秒和4秒后打印输出一段话。
注意到,在主线程写了
while 1:
pass
1
2
while 1:
pass
这是让主线程一直在等待
如果去掉上面两行,那就直接输出
Main Finished
1
Main Finished
程序执行结束。
使用Threading模块创建线程
使用Threading模块创建线程,直接从threading.Thread继承,然后重写init方法和run方法:
!/usr/bin/python
-- coding: UTF-8 --
import threading
import time
import thread
exitFlag = 0
class myThread (threading.Thread): #继承父类threading.Thread
def init(self, threadID, name, counter):
threading.Thread.init(self)
self.threadID = threadID
self.name = name
self.counter = counter
def run(self): #把要执行的代码写到run函数里面 线程在创建后会直接运行run函数
print "Starting " + self.name
print_time(self.name, self.counter, 5)
print "Exiting " + self.name
def print_time(threadName, delay, counter):
while counter:
if exitFlag:
thread.exit()
time.sleep(delay)
print "%s: %s" % (threadName, time.ctime(time.time()))
counter -= 1
创建新线程
thread1 = myThread(1, "Thread-1", 1)
thread2 = myThread(2, "Thread-2", 2)
开启线程
thread1.start()
thread2.start()
print "Exiting Main Thread"
!/usr/bin/python
-- coding: UTF-8 --
import threading
import time
import thread
exitFlag = 0
class myThread (threading.Thread): #继承父类threading.Thread
def __init__(self, threadID, name, counter):
threading.Thread.__init__(self)
self.threadID = threadID
self.name = name
self.counter = counter
def run(self): #把要执行的代码写到run函数里面 线程在创建后会直接运行run函数
print "Starting " + self.name
print_time(self.name, self.counter, 5)
print "Exiting " + self.name
def print_time(threadName, delay, counter):
while counter:
if exitFlag:
thread.exit()
time.sleep(delay)
print "%s: %s" % (threadName, time.ctime(time.time()))
counter -= 1
创建新线程
thread1 = myThread(1, "Thread-1", 1)
thread2 = myThread(2, "Thread-2", 2)
开启线程
thread1.start()
thread2.start()
print "Exiting Main Thread"
运行结果:
Starting Thread-1Starting Thread-2
Exiting Main Thread
Thread-1: Thu Nov 3 18:42:19 2016
Thread-2: Thu Nov 3 18:42:20 2016
Thread-1: Thu Nov 3 18:42:20 2016
Thread-1: Thu Nov 3 18:42:21 2016
Thread-2: Thu Nov 3 18:42:22 2016
Thread-1: Thu Nov 3 18:42:22 2016
Thread-1: Thu Nov 3 18:42:23 2016
Exiting Thread-1
Thread-2: Thu Nov 3 18:42:24 2016
Thread-2: Thu Nov 3 18:42:26 2016
Thread-2: Thu Nov 3 18:42:28 2016
Exiting Thread-2
Starting Thread-1Starting Thread-2
Exiting Main Thread
Thread-1: Thu Nov 3 18:42:19 2016
Thread-2: Thu Nov 3 18:42:20 2016
Thread-1: Thu Nov 3 18:42:20 2016
Thread-1: Thu Nov 3 18:42:21 2016
Thread-2: Thu Nov 3 18:42:22 2016
Thread-1: Thu Nov 3 18:42:22 2016
Thread-1: Thu Nov 3 18:42:23 2016
Exiting Thread-1
Thread-2: Thu Nov 3 18:42:24 2016
Thread-2: Thu Nov 3 18:42:26 2016
Thread-2: Thu Nov 3 18:42:28 2016
Exiting Thread-2
有没有发现什么奇怪的地方?打印的输出格式好奇怪。比如第一行之后应该是一个回车的,结果第二个进程就打印出来了。
那是因为什么?因为这几个线程没有设置同步。
线程同步
如果多个线程共同对某个数据修改,则可能出现不可预料的结果,为了保证数据的正确性,需要对多个线程进行同步。
使用Thread对象的Lock和Rlock可以实现简单的线程同步,这两个对象都有acquire方法和release方法,对于那些需要每次只允许一个线程操作的数据,可以将其操作放到acquire和release方法之间。如下:
多线程的优势在于可以同时运行多个任务(至少感觉起来是这样)。但是当线程需要共享数据时,可能存在数据不同步的问题。
考虑这样一种情况:一个列表里所有元素都是0,线程”set”从后向前把所有元素改成1,而线程”print”负责从前往后读取列表并打印。
那么,可能线程”set”开始改的时候,线程”print”便来打印列表了,输出就成了一半0一半1,这就是数据的不同步。为了避免这种情况,引入了锁的概念。
锁有两种状态——锁定和未锁定。每当一个线程比如”set”要访问共享数据时,必须先获得锁定;如果已经有别的线程比如”print”获得锁定了,那么就让线程”set”暂停,也就是同步阻塞;等到线程”print”访问完毕,释放锁以后,再让线程”set”继续。
经过这样的处理,打印列表时要么全部输出0,要么全部输出1,不会再出现一半0一半1的尴尬场面。
看下面的例子:
Python
-- coding: UTF-8 --
import threading
import time
class myThread (threading.Thread):
def init(self, threadID, name, counter):
threading.Thread.init(self)
self.threadID = threadID
self.name = name
self.counter = counter
def run(self):
print "Starting " + self.name
# 获得锁,成功获得锁定后返回True
# 可选的timeout参数不填时将一直阻塞直到获得锁定
# 否则超时后将返回False
threadLock.acquire()
print_time(self.name, self.counter, 3)
# 释放锁
threadLock.release()
def print_time(threadName, delay, counter):
while counter:
time.sleep(delay)
print "%s: %s" % (threadName, time.ctime(time.time()))
counter -= 1
threadLock = threading.Lock()
threads = []
创建新线程
thread1 = myThread(1, "Thread-1", 1)
thread2 = myThread(2, "Thread-2", 2)
开启新线程
thread1.start()
thread2.start()
添加线程到线程列表
threads.append(thread1)
threads.append(thread2)
等待所有线程完成
for t in threads:
t.join()
print "Exiting Main Thread"
-- coding: UTF-8 --
import threading
import time
class myThread (threading.Thread):
def __init__(self, threadID, name, counter):
threading.Thread.__init__(self)
self.threadID = threadID
self.name = name
self.counter = counter
def run(self):
print "Starting " + self.name
# 获得锁,成功获得锁定后返回True
# 可选的timeout参数不填时将一直阻塞直到获得锁定
# 否则超时后将返回False
threadLock.acquire()
print_time(self.name, self.counter, 3)
# 释放锁
threadLock.release()
def print_time(threadName, delay, counter):
while counter:
time.sleep(delay)
print "%s: %s" % (threadName, time.ctime(time.time()))
counter -= 1
threadLock = threading.Lock()
threads = []
创建新线程
thread1 = myThread(1, "Thread-1", 1)
thread2 = myThread(2, "Thread-2", 2)
开启新线程
thread1.start()
thread2.start()
添加线程到线程列表
threads.append(thread1)
threads.append(thread2)
等待所有线程完成
for t in threads:
t.join()
print "Exiting Main Thread"
在上面的代码中运用了线程锁还有join等待。
运行结果如下:
Starting Thread-1
Starting Thread-2
Thread-1: Thu Nov 3 18:56:49 2016
Thread-1: Thu Nov 3 18:56:50 2016
Thread-1: Thu Nov 3 18:56:51 2016
Thread-2: Thu Nov 3 18:56:53 2016
Thread-2: Thu Nov 3 18:56:55 2016
Thread-2: Thu Nov 3 18:56:57 2016
Exiting Main Thread
Starting Thread-1
Starting Thread-2
Thread-1: Thu Nov 3 18:56:49 2016
Thread-1: Thu Nov 3 18:56:50 2016
Thread-1: Thu Nov 3 18:56:51 2016
Thread-2: Thu Nov 3 18:56:53 2016
Thread-2: Thu Nov 3 18:56:55 2016
Thread-2: Thu Nov 3 18:56:57 2016
Exiting Main Thread
这样一来,你可以发现就不会出现刚才的输出混乱的结果了。
线程优先级队列
Python的Queue模块中提供了同步的、线程安全的队列类,包括FIFO(先入先出)队列Queue,LIFO(后入先出)队列LifoQueue,和优先级队列PriorityQueue。这些队列都实现了锁原语,能够在多线程中直接使用。可以使用队列来实现线程间的同步。
Queue模块中的常用方法:
Queue.qsize() 返回队列的大小
Queue.empty() 如果队列为空,返回True,反之False
Queue.full() 如果队列满了,返回True,反之False
Queue.full 与 maxsize 大小对应
Queue.get([block[, timeout]])获取队列,timeout等待时间
Queue.get_nowait() 相当Queue.get(False)
Queue.put(item) 写入队列,timeout等待时间
Queue.put_nowait(item) 相当Queue.put(item, False)
Queue.task_done() 在完成一项工作之后,Queue.task_done()函数向任务已经完成的队列发送一个信号
Queue.join() 实际上意味着等到队列为空,再执行别的操作
用一个实例感受一下:
-- coding: UTF-8 --
import Queue
import threading
import time
exitFlag = 0
class myThread (threading.Thread):
def init(self, threadID, name, q):
threading.Thread.init(self)
self.threadID = threadID
self.name = name
self.q = q
def run(self):
print "Starting " + self.name
process_data(self.name, self.q)
print "Exiting " + self.name
def process_data(threadName, q):
while not exitFlag:
queueLock.acquire()
if not workQueue.empty():
data = q.get()
queueLock.release()
print "%s processing %s" % (threadName, data)
else:
queueLock.release()
time.sleep(1)
threadList = ["Thread-1", "Thread-2", "Thread-3"]
nameList = ["One", "Two", "Three", "Four", "Five"]
queueLock = threading.Lock()
workQueue = Queue.Queue(10)
threads = []
threadID = 1
创建新线程
for tName in threadList:
thread = myThread(threadID, tName, workQueue)
thread.start()
threads.append(thread)
threadID += 1
填充队列
queueLock.acquire()
for word in nameList:
workQueue.put(word)
queueLock.release()
等待队列清空
while not workQueue.empty():
pass
通知线程是时候退出
exitFlag = 1
等待所有线程完成
for t in threads:
t.join()
print "Exiting Main Thread"
-- coding: UTF-8 --
import Queue
import threading
import time
exitFlag = 0
class myThread (threading.Thread):
def __init__(self, threadID, name, q):
threading.Thread.__init__(self)
self.threadID = threadID
self.name = name
self.q = q
def run(self):
print "Starting " + self.name
process_data(self.name, self.q)
print "Exiting " + self.name
def process_data(threadName, q):
while not exitFlag:
queueLock.acquire()
if not workQueue.empty():
data = q.get()
queueLock.release()
print "%s processing %s" % (threadName, data)
else:
queueLock.release()
time.sleep(1)
threadList = ["Thread-1", "Thread-2", "Thread-3"]
nameList = ["One", "Two", "Three", "Four", "Five"]
queueLock = threading.Lock()
workQueue = Queue.Queue(10)
threads = []
threadID = 1
创建新线程
for tName in threadList:
thread = myThread(threadID, tName, workQueue)
thread.start()
threads.append(thread)
threadID += 1
填充队列
queueLock.acquire()
for word in nameList:
workQueue.put(word)
queueLock.release()
等待队列清空
while not workQueue.empty():
pass
通知线程是时候退出
exitFlag = 1
等待所有线程完成
for t in threads:
t.join()
print "Exiting Main Thread"
运行结果:
Starting Thread-1
Starting Thread-2
Starting Thread-3
Thread-3 processing One
Thread-1 processing Two
Thread-2 processing Three
Thread-3 processing Four
Thread-2 processing Five
Exiting Thread-2
Exiting Thread-3
Exiting Thread-1
Exiting Main Thread
1
2
3
4
5
6
7
8
9
10
11
12
Starting Thread-1
Starting Thread-2
Starting Thread-3
Thread-3 processing One
Thread-1 processing Two
Thread-2 processing Three
Thread-3 processing Four
Thread-2 processing Five
Exiting Thread-2
Exiting Thread-3
Exiting Thread-1
Exiting Main Thread