Python Language
Processer och trådar

Introduktion

Global tolklås

Python multithreading-prestanda kan ofta drabbas på grund av det globala tolklåset . Kort sagt, även om du kan ha flera trådar i ett Python-program, kan bara en bytekodinstruktion exekveras parallellt när som helst, oavsett antalet CPU: er.

Som sådan kan flertrådning i fall där operationer blockeras av externa händelser - som nätverksåtkomst - vara ganska effektiv:

import threading
import time


def process():
    time.sleep(2)


start = time.time()
process()
print("One run took %.2fs" % (time.time() - start))


start = time.time()
threads = [threading.Thread(target=process) for _ in range(4)]
for t in threads:
    t.start()
for t in threads:
    t.join()
print("Four runs took %.2fs" % (time.time() - start))

# Out: One run took 2.00s
# Out: Four runs took 2.00s

Observera att även om varje process tog 2 sekunder att utföra, kunde de fyra processerna tillsammans fungera parallellt, vilket tog totalt 2 sekunder.

Men flertrådning i fall där intensiva beräkningar görs i Python-kod - till exempel mycket beräkning - resulterar inte i mycket förbättringar och kan till och med vara långsammare än att köra parallellt:

import threading
import time


def somefunc(i):
    return i * i

def otherfunc(m, i):
    return m + i

def process():
    for j in range(100):
        result = 0
        for i in range(100000):
            result = otherfunc(result, somefunc(i))


start = time.time()
process()
print("One run took %.2fs" % (time.time() - start))


start = time.time()
threads = [threading.Thread(target=process) for _ in range(4)]
for t in threads:
    t.start()
for t in threads:
    t.join()
print("Four runs took %.2fs" % (time.time() - start))

# Out: One run took 2.05s
# Out: Four runs took 14.42s

I det senare fallet kan multiprocessering vara effektiv eftersom flera processer naturligtvis kan utföra flera instruktioner samtidigt:

import multiprocessing
import time


def somefunc(i):
    return i * i

def otherfunc(m, i):
    return m + i

def process():
    for j in range(100):
        result = 0
        for i in range(100000):
            result = otherfunc(result, somefunc(i))


start = time.time()
process()
print("One run took %.2fs" % (time.time() - start))


start = time.time()
processes = [multiprocessing.Process(target=process) for _ in range(4)]
for p in processes:
    p.start()
for p in processes:
    p.join()
print("Four runs took %.2fs" % (time.time() - start))

# Out: One run took 2.07s
# Out: Four runs took 2.30s

Kör i flera trådar

Använd threading.Thread att köra en funktion i en annan tråd.

import threading
import os

def process():
    print("Pid is %s, thread id is %s" % (os.getpid(), threading.current_thread().name))

threads = [threading.Thread(target=process) for _ in range(4)]
for t in threads:
    t.start()
for t in threads:
    t.join()
    
# Out: Pid is 11240, thread id is Thread-1
# Out: Pid is 11240, thread id is Thread-2
# Out: Pid is 11240, thread id is Thread-3
# Out: Pid is 11240, thread id is Thread-4

Kör i flera processer

Använd multiprocessing.Process att köra en funktion i en annan process. Gränssnittet liknar threading.Thread :

import multiprocessing
import os

def process():
    print("Pid is %s" % (os.getpid(),))

processes = [multiprocessing.Process(target=process) for _ in range(4)]
for p in processes:
    p.start()
for p in processes:
    p.join()
    
# Out: Pid is 11206
# Out: Pid is 11207
# Out: Pid is 11208
# Out: Pid is 11209

Dela tillstånd mellan trådar

Eftersom alla trådar körs i samma process har alla trådar tillgång till samma data.

Samtidig åtkomst till delad data bör dock skyddas med ett lås för att undvika synkroniseringsproblem.

import threading

obj = {}
obj_lock = threading.Lock()

def objify(key, val):
    print("Obj has %d values" % len(obj))
    with obj_lock:
        obj[key] = val
    print("Obj now has %d values" % len(obj))

ts = [threading.Thread(target=objify, args=(str(n), n)) for n in range(4)]
for t in ts:
    t.start()
for t in ts:
    t.join()
print("Obj final result:")
import pprint; pprint.pprint(obj)

# Out: Obj has 0 values
# Out:  Obj has 0 values
# Out: Obj now has 1 values
# Out: Obj now has 2 valuesObj has 2 values
# Out: Obj now has 3 values
# Out: 
# Out:  Obj has 3 values
# Out: Obj now has 4 values
# Out: Obj final result:
# Out: {'0': 0, '1': 1, '2': 2, '3': 3}

Dela tillstånd mellan processer

Kod som körs i olika processer delar inte som standard samma data. Men multiprocessing modulen innehåller primitiv för att dela värden över flera processer.

import multiprocessing

plain_num = 0
shared_num = multiprocessing.Value('d', 0)
lock = multiprocessing.Lock()

def increment():
    global plain_num
    with lock:
        # ordinary variable modifications are not visible across processes
        plain_num += 1
        # multiprocessing.Value modifications are
        shared_num.value += 1

ps = [multiprocessing.Process(target=increment) for n in range(4)]
for p in ps:
    p.start()
for p in ps:
    p.join()

print("plain_num is %d, shared_num is %d" % (plain_num, shared_num.value))

# Out: plain_num is 0, shared_num is 4