在看这篇文章之前请先看
本文采取例子进行循序渐进。
多线程无锁
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
| import asyncio
from concurrent.futures import ThreadPoolExecutor, as_completed
import time
import threading
import random
lock = threading.Lock()
num = 0
def add():
global num
for i in range(1000000):
num += 1
def sub():
global num
for i in range(1000000):
num -= 1
if __name__ == '__main__':
sub1 = threading.Thread(target=add)
sub2 = threading.Thread(target=sub)
sub1.start()
sub2.start()
sub1.join()
sub2.join()
print(num)
|
每次输出都不一样
多线程有锁
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
| import asyncio
from concurrent.futures import ThreadPoolExecutor, as_completed
import time
import threading
import random
lock = threading.Lock()
num = 0
def add():
global num
for i in range(1000000):
lock.acquire()
num += 1
lock.release()
def sub():
global num
for i in range(1000000):
lock.acquire()
num -= 1
lock.release()
if __name__ == '__main__':
sub1 = threading.Thread(target=add)
sub2 = threading.Thread(target=sub)
sub1.start()
sub2.start()
sub1.join()
sub2.join()
print(num)
|
输出 0
线程池无锁
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
| import asyncio
from concurrent.futures import ThreadPoolExecutor, as_completed
import time
import threading
import random
lock = threading.Lock()
num = 0
def add():
global num
for i in range(1000000):
num += 1
def sub():
global num
for i in range(1000000):
num -= 1
if __name__ == '__main__':
task_list = []
pool = ThreadPoolExecutor(max_workers=2)
task = pool.submit(add)
task_list.append(task)
task2 = pool.submit(sub)
task_list.append(task2)
print(num)
|
每次输出都不一样。
线程池有锁
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
| import asyncio
from concurrent.futures import ThreadPoolExecutor, as_completed
import time
import threading
import random
lock = threading.Lock()
num = 0
def add():
global num
for i in range(1000000):
lock.acquire()
num += 1
lock.release()
def sub():
global num
for i in range(1000000):
lock.acquire()
num -= 1
lock.release()
if __name__ == '__main__':
task_list = []
pool = ThreadPoolExecutor(max_workers=2)
task = pool.submit(add)
task_list.append(task)
task2 = pool.submit(sub)
task_list.append(task2)
print(num)
|
发现输出也不是 0。
这是因为，多线程在运行的时候，主线程没有等待多线程运行完就直接运行了主线程，所以，我们要加一个判断，等多线程运行完了，再运行主线程。
线程池有锁 优化
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
| import asyncio
from concurrent.futures import ThreadPoolExecutor, as_completed
import time
import threading
import random
lock = threading.Lock()
num = 0
def add():
global num
for i in range(1000000):
lock.acquire()
num += 1
lock.release()
def sub():
global num
for i in range(1000000):
lock.acquire()
num -= 1
lock.release()
if __name__ == '__main__':
task_list = []
pool = ThreadPoolExecutor(max_workers=2)
task = pool.submit(add)
task_list.append(task)
task2 = pool.submit(sub)
task_list.append(task2)
for result in as_completed(task_list):
data = result.result()
print(num)
|
输出 0。
