python experiment with gc and memory_profiler

Question

I ran the following with python 2.7.7:

import gc
import memory_profiler

print memory_profiler.memory_usage()[0]
x = [i for i in range(10000000)]
print memory_profiler.memory_usage()[0]
x = None
print memory_profiler.memory_usage()[0]
gc.collect()
print memory_profiler.memory_usage()[0]

It produced the following output:

11.36328125
321.9140625
245.6171875
245.6171875

What is going on here? What explains the ~25% reduction in memory usage?

You have a big list that you delete with x=None. Nowever all the int objects that you allocated (the i) are in a cache of small objects that is never freed on the basis that you will reuse them. — Barry Scott, May 06 '16 at 17:09

score 1 · Answer 1 · edited May 23 '17 at 11:52

This post should give you a very good idea. Basically, the python will have some free lists memory and when they are used up the memory is then overhead. For example:

import gc
import memory_profiler

print memory_profiler.memory_usage()[0]
x = [i for i in range(10000)]
print memory_profiler.memory_usage()[0]
x = None
print memory_profiler.memory_usage()[0]
gc.collect()
print memory_profiler.memory_usage()[0]

Output:

But when I ran your code with a really huge list, the results were different, code:

import gc
import memory_profiler


print memory_profiler.memory_usage()[0]
x = [i for i in range(10000000)]
print memory_profiler.memory_usage()[0]
x = None
print memory_profiler.memory_usage()[0]
gc.collect()
print memory_profiler.memory_usage()[0]

Output:

7.3515625
387.31640625
311.30859375
94.7890625

So if everything I said is true, if it is truely causing overhead after eating up free python list memory; let's try releasing the memory similar to this post:

import gc
import memory_profiler

def release_list(a):
   del a[:]
   del a


print memory_profiler.memory_usage()[0]
x = [i for i in range(10000000)]
release_list(x)
print memory_profiler.memory_usage()[0]
x = None
print memory_profiler.memory_usage()[0]
gc.collect()
print memory_profiler.memory_usage()[0]

Output:

7.34765625
318.3359375
318.3359375
96.3359375

Clearly, when you assign x = None it releases the extra burden you have initiated with really big list. Usually in real world the python free list memory should suffice users requirements and would make no difference.

Additional resources:

python experiment with gc and memory_profiler

1 Answers1