Develop Reference

Why do python Processes behave differently with barrier.wait()?

TLDR
Selecting the process that gets 0 back from barrier.wait() biases the order of the processes on the next iteration. Why?
Full Story
The below problem is a simple version of my actual problem.
Lets say I have the following multiprocessing problem:
I have a string (named output), n processes, n characters (not spaces), and N iterations.
Example: output, n, characters, N = '', 3, ['a', 'b', 'c'], 4
For each iteration, every process should append its character to the string (in any order). After each iteration, a space character should be appended to the string.
For the example, the output could be:
output = 'bca abc bac cab'
To get this functionality I use the multiprocessing library.
from multiprocessing import Process, Lock, Value, Manager, Barrier
import ctypes
def print_characters(lock, barrier, character, N, output):
for i in range(N):
lock.acquire()
output.value += character
lock.release()
_id = barrier.wait() # get an id for each thread (although not promised to be assigned randomly)
if _id == 0: # select 1 and get this thread to append the ' '
lock.acquire()
output.value += ' '
lock.release()
barrier.wait()
if __name__ == '__main__':
manager = Manager()
output = manager.Value(ctypes.c_char_p, "")
num_processes = 3
characters = 'abc'
N = 4
lock, barrier = Lock(), Barrier(num_processes)
processes = []
for i in range(num_processes):
p = Process(target=print_characters, args=(lock, barrier, characters[i], N, output))
p.start()
processes.append(p)
for p in processes:
p.join()
print(output.value)
Cool....
According to the documentation for Barrier.wait(),
When all the [processes] to the barrier have called this function, they are all released simultaneously.
So, if all n processes are released simultaneously, I wondered if this would cause a kind of "uniform random race condition" on the processes such that each process had an equal chance of getting to the lock.acquire() on the next iteration.
I don't really care about the order but it might be nice to know if I have the processes doing something different in the future.
To see this if this was the case I counted the number of times that each character came in which position:
output = output.value.split(' ')[:-1]
c = [[0]*num_processes for i in range(num_processes)]
for line in output:
for i, character in enumerate(line):
ind = characters.find(character)
c[ind][i] += 1
print(c)
For N = 1000, this is one output I got:
[[1000 0 0]
[ 0 948 52]
[ 0 52 948]]
:| ok?
So out of 1000 iterations, the process with the first character was always first. That didn't seem very random.
I then made one change in the print_characters function:
if _id == 1:
And I got:
[[254 254 492]
[489 489 22]
[257 257 486]]
Not uniform random but also not as biased.
Changing the number of processes (assuming the cpu can support it) and thus the number of characters shows this same effect:
Selecting the process that gets 0 back from barrier.wait() biases the order of the processes on the next iteration.
Why?

How to maintain global processes in a pool working recursively?

I want to implement a recursive parallel algorithm and I want a pool to be created only once and each time step do a job wait for all the jobs to finish and then call the processes again with inputs the previous outputs and then again the same at the next time step, etc.
My problem is that I have implemented a version where every time step I create and kill the pool, but this is extremely slow, even slower than the sequential version. When I try to implement a version where the pool is created only once at the beginning I got assertion error when I try to call join().
This is my code
def log_result(result):
tempx , tempb, u = result
X[:,u,np.newaxis], b[:,u,np.newaxis] = tempx , tempb
workers = mp.Pool(processes = 4)
for t in range(p,T):
count = 0 #==========This is only master's job=============
for l in range(p):
for k in range(4):
gn[count]=train[t-l-1,k]
count+=1
G = G*v + gn # gn.T#==================================
if __name__ == '__main__':
for i in range(4):
workers.apply_async(OULtraining, args=(train[t,i], X[:,i,np.newaxis], b[:,i,np.newaxis], i, gn), callback = log_result)
workers.join()
X and b are the matrices that I want to update directly at the master's memory.
What is wrong here and I get the assertion error?
Can I implement with the pool what I want or not?

You cannot join a pool that is not closed first, as join() will wait worker processes to terminate, not jobs to complete (https://docs.python.org/3.6/library/multiprocessing.html section 17.2.2.9).
But as this will close the pool, which is not what you want, you cannot use this. So join is out, and you need to implement a "wait until all jobs completed" by yourself.
One way of doing this without busy loops would be using a queue. You could also work with bounded semaphores, but they do not work on all operating systems.
counter = 0
lock_queue = multiprocessing.Queue()
counter_lock = multiprocessing.Lock()
def log_result(result):
tempx , tempb, u = result
X[:,u,np.newaxis], b[:,u,np.newaxis] = tempx , tempb
with counter_lock:
counter += 1
if counter == 4:
counter = 0
lock_queue.put(42)
workers = mp.Pool(processes = 4)
for t in range(p,T):
count = 0 #==========This is only master's job=============
for l in range(p):
for k in range(4):
gn[count]=train[t-l-1,k]
count+=1
G = G*v + gn # gn.T#==================================
if __name__ == '__main__':
counter = 0
for i in range(4):
workers.apply_async(OULtraining, args=(train[t,i], X[:,i,np.newaxis], b[:,i,np.newaxis], i, gn), callback = log_result)
lock_queue.get(block=True)
This resets a global counter before submitting jobs. As soon as a job is completed, you callback increments a global counter. When the counter hits 4 (your number of jobs), the callback knows it has processed the last result. Then a dummy message is sent in a queue. Your main program is waiting at Queue.get() for something to appear there.
This allows your main program to block until all jobs have completed, without closing down the pool.
If you replace multiprocessing.Pool with ProcessPoolExecutor from concurrent.futures, you can skip this part and use
concurrent.futures.wait(fs, timeout=None, return_when=ALL_COMPLETED)
to block until all submitted tasks have finished. From functional standpoint there is no difference between these. The concurrent.futures method is a couple of lines shorter but the result is exactly the same.

Can Python threads work on the same process?

I am trying to come up with a way to have threads work on the same goal without interfering. In this case I am using 4 threads to add up every number between 0 and 90,000. This code runs but it ends almost immediately (Runtime: 0.00399994850159 sec) and only outputs 0. Originally I wanted to do it with a global variable but I was worried about the threads interfering with each other (ie. the small chance that two threads double count or skip a number due to strange timing of the reads/writes). So instead I distributed the workload beforehand. If there is a better way to do this please share. This is my simple way of trying to get some experience into multi threading. Thanks
import threading
import time
start_time = time.time()
tot1 = 0
tot2 = 0
tot3 = 0
tot4 = 0
def Func(x,y,tot):
tot = 0
i = y-x
while z in range(0,i):
tot = tot + i + z
# class Tester(threading.Thread):
# def run(self):
# print(n)
w = threading.Thread(target=Func, args=(0,22499,tot1))
x = threading.Thread(target=Func, args=(22500,44999,tot2))
y = threading.Thread(target=Func, args=(45000,67499,tot3))
z = threading.Thread(target=Func, args=(67500,89999,tot4))
w.start()
x.start()
y.start()
z.start()
w.join()
x.join()
y.join()
z.join()
# while (w.isAlive() == False | x.isAlive() == False | y.isAlive() == False | z.isAlive() == False): {}
total = tot1 + tot2 + tot3 + tot4
print total
print("--- %s seconds ---" % (time.time() - start_time))

You have a bug that makes this program end almost immediately. Look at while z in range(0,i): in Func. z isn't defined in the function and its only by luck (bad luck really) that you happen to have a global variable z = threading.Thread(target=Func, args=(67500,89999,tot4)) that masks the problem. You are testing whether the thread object is in a list of integers... and its not!
The next problem is with the global variables. First, you are absolutely right that using a single global variable is not thread safe. The threads would mess with each others calculations. But you misunderstand how globals work. When you do threading.Thread(target=Func, args=(67500,89999,tot4)), python passes the object currently referenced by tot4 to the function, but the function has no idea which global it came from. You only update the local variable tot and discard it when the function completes.
A solution is to use a global container to hold the calculations as shown in the example below. Unfortunately, this is actually slower than just doing all the work in one thread. The python global interpreter lock (GIL) only lets 1 thread run at a time and only slows down CPU-intensive tasks implemented in pure python.
You could look at the multiprocessing module to split this into multiple processes. That works well if the cost of running the calculation is large compared to the cost of starting the process and passing it data.
Here is a working copy of your example:
import threading
import time
start_time = time.time()
tot = [0] * 4
def Func(x,y,tot_index):
my_total = 0
i = y-x
for z in range(0,i):
my_total = my_total + i + z
tot[tot_index] = my_total
# class Tester(threading.Thread):
# def run(self):
# print(n)
w = threading.Thread(target=Func, args=(0,22499,0))
x = threading.Thread(target=Func, args=(22500,44999,1))
y = threading.Thread(target=Func, args=(45000,67499,2))
z = threading.Thread(target=Func, args=(67500,89999,3))
w.start()
x.start()
y.start()
z.start()
w.join()
x.join()
y.join()
z.join()
# while (w.isAlive() == False | x.isAlive() == False | y.isAlive() == False | z.isAlive() == False): {}
total = sum(tot)
print total
print("--- %s seconds ---" % (time.time() - start_time))

You can pass in a mutable object that you can add your results either with an identifier, e.g. dict or just a list and append() the results, e.g.:
import threading
def Func(start, stop, results):
results.append(sum(range(start, stop+1)))
rngs = [(0, 22499), (22500, 44999), (45000, 67499), (67500, 89999)]
results = []
jobs = [threading.Thread(target=Func, args=(start, stop, results)) for start, stop in rngs]
for j in jobs:
j.start()
for j in jobs:
j.join()
print(sum(results))
# 4049955000
# 100 loops, best of 3: 2.35 ms per loop

As others have noted you could look multiprocessing in order to split the work to multiple different processes that can run parallel. This would benefit especially in CPU-intensive tasks assuming that there isn't huge amount of data to pass between the processes.
Here's a simple implementation of the same functionality using multiprocessing:
from multiprocessing import Pool
POOL_SIZE = 4
NUMBERS = 90000
def func(_range):
tot = 0
for z in range(*_range):
tot += z
return tot
with Pool(POOL_SIZE) as pool:
chunk_size = int(NUMBERS / POOL_SIZE)
chunks = ((i, i + chunk_size) for i in range(0, NUMBERS, chunk_size))
print(sum(pool.imap(func, chunks)))
In above chunks is a generator that produces the same ranges that were hardcoded in original version. It's given to imap which works the same as standard map except that it executes the function in the processes within the pool.
Less known fact about multiprocessing is that you can easily convert the code to use threads instead of processes by using undocumented multiprocessing.pool.ThreadPool. In order to convert above example to use threads just change import to:
from multiprocessing.pool import ThreadPool as Pool

Python threads don't run simultaneously

I'm brand new to multi-threaded processing, so please forgive me if I butcher terms or miss something obvious.
The code below doesn't offer any time advantage over different code that calls the same two functions one after the other.
import time
import threading
start_time = time.clock()
def fibonacci(nth): #can be ignored
first = 0
second = 1
for i in range(nth):
third = first + second
first = second
second = third
print "Fibonacci number", i + 1, "is", len(str(first)), "digits long"
def collatz(collatz_max): #can be ignored
for n in range(collatz_max):
n = n + 1 #avoid entering 0
solution = []
solution.append(n)
while n != 1:
if n % 2 == 0:
n = n / 2
else:
n = (n*3) + 1
solution.append(n)
print "Path for Collatz number", collatz_max, "is", solution
def scripts():
thread_fibonacci = threading.Thread(target=fibonacci, args = (800000,))
thread_collatz = threading.Thread(target=collatz, args = (400000,))
thread_fibonacci.start()
thread_collatz.start()
return thread_fibonacci, thread_collatz
all_scripts = scripts()
#wait until both threads are finished
for script in all_scripts:
script.join()
print time.clock() - start_time, "seconds"
What do I need to do to make the threads simultaneous? Does GIL mean concurrency can only be achieved through separate processes? If so, what is the point of multithreading?
Using Python 2.7.5 on Windows 8.1, quad-core processor. Any help would be appreciated.

There are good answers regarding the GIL you can look at.
In short, if your tasks are CPU-bound (like the ones you posted), threads are not going to help you. Python threads are good for IO-bound tasks, like retrieving a web page.

Multiprocessing in python to speed up functions

I am confused with Python multiprocessing.
I am trying to speed up a function which process strings from a database but I must have misunderstood how multiprocessing works because the function takes longer when given to a pool of workers than with “normal processing”.
Here an example of what I am trying to achieve.
from time import clock, time
from multiprocessing import Pool, freeze_support
from random import choice
def foo(x):
TupWerteMany = []
for i in range(0,len(x)):
TupWerte = []
s = list(x[i][3])
NewValue = choice(s)+choice(s)+choice(s)+choice(s)
TupWerte.append(NewValue)
TupWerte = tuple(TupWerte)
TupWerteMany.append(TupWerte)
return TupWerteMany
if __name__ == '__main__':
start_time = time()
List = [(u'1', u'aa', u'Jacob', u'Emily'),
(u'2', u'bb', u'Ethan', u'Kayla')]
List1 = List*1000000
# METHOD 1 : NORMAL (takes 20 seconds)
x2 = foo(List1)
print x2[1:3]
# METHOD 2 : APPLY_ASYNC (takes 28 seconds)
# pool = Pool(4)
# Werte = pool.apply_async(foo, args=(List1,))
# x2 = Werte.get()
# print '--------'
# print x2[1:3]
# print '--------'
# METHOD 3: MAP (!! DOES NOT WORK !!)
# pool = Pool(4)
# Werte = pool.map(foo, args=(List1,))
# x2 = Werte.get()
# print '--------'
# print x2[1:3]
# print '--------'
print 'Time Elaspse: ', time() - start_time
My questions:
Why does apply_async takes longer than the “normal way” ?
What I am doing wrong with map?
Does it makes sense to speed up such tasks with multiprocessing at all?
Finally: after all I have read here, I am wondering if multiprocessing in python works on windows at all ?

So your first problem is that there is no actual parallelism happening in foo(x), you are passing the entire list to the function once.
1)
The idea of a process pool is to have many processes doing computations on separate bits of some data.
# METHOD 2 : APPLY_ASYNC
jobs = 4
size = len(List1)
pool = Pool(4)
results = []
# split the list into 4 equally sized chunks and submit those to the pool
heads = range(size/jobs, size, size/jobs) + [size]
tails = range(0,size,size/jobs)
for tail,head in zip(tails, heads):
werte = pool.apply_async(foo, args=(List1[tail:head],))
results.append(werte)
pool.close()
pool.join() # wait for the pool to be done
for result in results:
werte = result.get() # get the return value from the sub jobs
This will only give you an actual speedup if the time it takes to process each chunk is greater than the time it takes to launch the process, in the case of four processes and four jobs to be done, of course these dynamics change if you've got 4 processes and 100 jobs to be done. Remember that you are creating a completely new python interpreter four times, this isn't free.
2) The problem you have with map is that it applies foo to EVERY element in List1 in a separate process, this will take quite a while. So if you're pool has 4 processes map will pop an item of the list four times and send it to a process to be dealt with - wait for process to finish - pop some more stuff of the list - wait for the process to finish. This makes sense only if processing a single item takes a long time, like for instance if every item is a file name pointing to a one gigabyte text file. But as it stands map will just take a single string of the list and pass it to foo where as apply_async takes a slice of the list. Try the following code
def foo(thing):
print thing
map(foo, ['a','b','c','d'])
That's the built-in python map and will run a single process, but the idea is exactly the same for the multiprocess version.
Added as per J.F.Sebastian's comment: You can however use the chunksize argument to map to specify an approximate size of for each chunk.
pool.map(foo, List1, chunksize=size/jobs)
I don't know though if there is a problem with map on Windows as I don't have one available for testing.
3) yes, given that your problem is big enough to justify forking out new python interpreters
4) can't give you a definitive answer on that as it depends on the number of cores/processors etc. but in general it should be fine on Windows.

On question (2)
With the guidance of Dougal and Matti, I figured out what's went wrong.
The original foo function processes a list of lists, while map requires a function to process single elements.
The new function should be
def foo2 (x):
TupWerte = []
s = list(x[3])
NewValue = choice(s)+choice(s)+choice(s)+choice(s)
TupWerte.append(NewValue)
TupWerte = tuple(TupWerte)
return TupWerte
and the block to call it :
jobs = 4
size = len(List1)
pool = Pool()
#Werte = pool.map(foo2, List1, chunksize=size/jobs)
Werte = pool.map(foo2, List1)
pool.close()
print Werte[1:3]
Thanks to all of you who helped me understand this.
Results of all methods:
for List * 2 Mio records: normal 13.3 seconds , parallel with async: 7.5 seconds, parallel with with map with chuncksize : 7.3, without chunksize 5.2 seconds

Here's a generic multiprocessing template if you are interested.
import multiprocessing as mp
import time
def worker(x):
time.sleep(0.2)
print "x= %s, x squared = %s" % (x, x*x)
return x*x
def apply_async():
pool = mp.Pool()
for i in range(100):
pool.apply_async(worker, args = (i, ))
pool.close()
pool.join()
if __name__ == '__main__':
apply_async()
And the output looks like this:
x= 0, x squared = 0
x= 1, x squared = 1
x= 2, x squared = 4
x= 3, x squared = 9
x= 4, x squared = 16
x= 6, x squared = 36
x= 5, x squared = 25
x= 7, x squared = 49
x= 8, x squared = 64
x= 10, x squared = 100
x= 11, x squared = 121
x= 9, x squared = 81
x= 12, x squared = 144
As you can see, the numbers are not in order, as they are being executed asynchronously.