RememberMe is a handy tool for memory problems in Python. It computes the total memory usage of Python objects.

sys.getsizeof is almost confusing in Python:

import sys
a = [1, 2, 3]
b = [a, a, a]
print(sys.getsizeof(a) == sys.getsizeof(b))  # Can you believe the result is `True`?

While rememberme gives you a clear idea how large an object is.

from rememberme import memory
a = [1, 2, 3]
b = [a, a, a]
print(memory(a))  # 172 bytes!
print(memory(b))  # 260 bytes!

pip install rememberme

Check out memory usage in the current frame:

from rememberme import memory
def foo():
    a = [1, 2, 3]
    b = [a, a, a]
    print(memory())
foo()  # 260 bytes. Note `a` is included in `b`.

Check out top memory consumers:

from rememberme import top
def foo():
    a = [1, 2, 3]
    b = [a, a, a]
    mem_top = top()  # with no args, check current frame
    print(mem_top[0])  # `b` and its memory usage
    print(mem_top[1])  # `a` and its memory usage

Even pretty print the result!

from rememberme import mem_print
def foo():
    a = [1, 2, 3]
    b = [a, a, a]
    mem_print(b)
foo()

Output:

                           ┌int (28.0B)
             ┌list (172.0B)┼int (28.0B)
             │             └int (28.0B)
             │             ┌int (28.0B)
list (260.0B)┼list (172.0B)┼int (28.0B)
             │             └int (28.0B)
             │             ┌int (28.0B)
             └list (172.0B)┼int (28.0B)
                           └int (28.0B)

• For better performance (and making better sense), the global dict, as well as modules, are not included in the memory usage of any objects.
• We essentially relies on tp_traverse to traverse the object graph. For C extensions, memory usage might be underestimated under various circumstances. For the most common numpy.ndarray, a specific procedure is defined to probe the memory usage correctly, but no correctness is guaranteed for other C extensions, which may have undetectable momery leaks within themselves.