lnever/pymemory

calculate memory footprint of python objects

pymemory

inspect memory footprint of python objects.

install

git clone github.com/mynameisvinn/pymemory
cd pymemory
python setup.py install

example

for example, we'd like to determine size of int 1. since a python object consumes 16 bytes of overhead and each int is 1 byte, so we should expect 24 bytes for 1.

>>> from pymemory import deep_getsizeof
>>> x = 1
>>> deep_getsizeof(x, set())  # prints 24 bytes

a deeper dive into python internals

in the above example, a cython struct PyLongObject was created to hold the integer 1.

a PyLongObject struct holds three attributes: a PyObject_HEAD object (16 bytes); length of array (4 bytes); and an array for value (another 4 bytes).

the PyObject_HEAD is a struct:

// https://code.woboq.org/llvm/include/python2.7/object.h.html

/* PyObject_HEAD defines the initial segment of every PyObject. */
PyObject_HEAD
{
	size_t refcnt;  // 8 bytes
	typeobject *type;  // 8 bytes
}