Learn Python – Python Memory Management- Basic and advance

In this tutorial, we will examine how Python manages the memory or how Python handles our date internally. We will dive deep into this subject matter to understand inner working of Python and how it handles the memory.

This tutorial will provide a deep perception of Python memory management. When we execute our Python script, there are so many common sense runs at the back of in Python memory to make the code efficient.


Memory administration is very necessary for software developers to work successfully with any programming language. As we know, Python is a famous and broadly used programming language. It is used nearly in each technical domain. In contrast to a programming language, memory administration is associated to writing memory-efficient code. We cannot overlook the importance of memory administration whilst implementing a large quantity of data. Improper memory management leads to slowness on the software and the server-side components. It also will become the motive of unsuitable working. If the memory is not treated well, it will take a good deal time while preprocessing the data.

In Python, reminiscence is managed via the Python supervisor which determines the place to put the utility information in the memory. So, we must have the information of Python reminiscence manager to write efficient code and maintainable code.

Let’s expect reminiscence appears like an empty book and we desire to write anything on the book’s page. Then, we write statistics any information the supervisor discover the free space in the e book and provide it to the application. The method of offering reminiscence to objects is referred to as allocation.

On the different side, when records is no longer use, it can be deleted by using the Python reminiscence manager. But the query is, how? And where did this memory come from?

Python Memory Allocation

Memory allocation is an critical phase of the memory management for a developer. This technique essentially allots free space in the computer’s virtual memory, and there are two types of digital memory works whilst executing programs.

Static Memory Allocation

Dynamic Memory Allocation

Static Memory Allocation –

Static memory allocation occurs at the compile time. For example – In C/C++, we declare a static array with the fixed sizes. Memory is allocated at the time of compilation. However, we cannot use the reminiscence again in the similarly program.

static int a=10;  

Stack Allocation

The Stack data shape is used to store the static memory. It is only wished interior the specific characteristic or approach call. The feature is added in program’s name stack on every occasion we name it. Variable challenge internal the characteristic is briefly stored in the feature call stack; the feature returns the value, and the name stack strikes to the text task. The compiler handles all these processes, so we don’t want to worry about it.

Call stack (stack facts structure) holds the program’s operational information such as subroutines or feature name in the order they are to be called. These functions are popped up from the stack when we called.

Dynamic Memory Allocation

Unlike static reminiscence allocation, Dynamic memory allocates the reminiscence at the runtime to the program. For instance – In C/C++, there is a predefined size of the integer of go with the flow records kind however there is no predefine size of the information types. Memory is allocated to the objects at the run time. We use the Heap for implement dynamic memory management. We can use the reminiscence in the course of the program.

int *a;  
p = new int;  

As we know, everything in Python is an object skill dynamic reminiscence allocation inspires the Python reminiscence management. Python reminiscence supervisor automatically vanishes when the object is no longer in use.

Heap Memory Allocation

Heap statistics structure is used for dynamic reminiscence which is now not related to naming counterparts. It is type of reminiscence that makes use of outside the program at the world space. One of the satisfactory blessings of heap memory is to it freed up the memory space if the object is no longer in use or the node is deleted.

In the under example, we define how the function’s variable save in the stack and a heap.

Default Python Implementation

Python is an open-source, object-oriented programming language which default carried out in the C programming language. That’s very fascinating fact – A language which is most popular written in every other language? But this is now not a entire truth, but kind of.

Basically, Python language is written in the English language. However, it is described in the reference manual that isn’t always beneficial with the aid of itself. So, we need an interpreter primarily based code on the rule in the manual.

The gain of the default implementation, it executes the Python code in the laptop and it additionally converts our Python code into instruction. So, we can say that Python’s default implementation fulfills the each requirements.

Note – Virtual Machines are now not the bodily computer, however they are instigated in the software.

The program that we write the usage of Python language first converts into the computer-relatable guidelines bytecode. The virtual laptop interprets this bytecode.

Python Garbage Collector

As we have explained earlier, Python eliminates those objects that are no longer in use or can say that it frees up the memory space. This process of vanish the pointless object’s memory area is called the Garbage Collector. The Python rubbish collector initiates its execution with the program and is activated if the reference count falls to zero.

When we assign the new name or placed it in containers such as a dictionary or tuple, the reference count number increases its value. If we reassign the reference to an object, the reference counts decreases its value if. It also decreases its cost when the object’s reference goes out of scope or an object is deleted.

As we know, Python makes use of the dynamic memory allocation which is managed by the Heap data structure. Memory Heap holds the objects and other records buildings that will be used in the program. Python memory manager manages the allocation or de-allocation of the heap reminiscence house thru the API functions.

Python Objects in Memory

As we know, the entirety in Python is object. The object can both be simple (containing numbers, strings, etc.) or containers (dictionary, lists, or person defined classes). In Python, we do not need to declare the variables or their kinds before the usage of them in a program.

Let’s understand the following example.

Example –

a= 10  
 del a  


Traceback (most recent call last):
  File "", line 1, in 
NameError : name 'a' is not defined

As we can see in the above output, we assigned the value to object x and printed it. When we put off the object x and attempt to get right of entry to in further code, there will be an error that claims that the variable x is no longer defined.

Hence, Python garbage collector works automatically and the programmers would not need to fear about it, in contrast to C.

Reference Counting in Python

Reference counting states that how many instances other objects reference an object. When a reference of the object is assigned, the matter of object is incremented one. When references of an object are eliminated or deleted, the count of object is decremented. The Python reminiscence supervisor performs the de-allocation when the reference depend turns into zero. Let’s make it easy to understand.

Example –

Suppose, there is two or greater variable that incorporates the identical value, so the Python digital laptop alternatively creating some other object of the same price in the private heap. It really makes the second variable point to that the initially existing fee in the private heap.

This is highly advisable to retain the memory, which can be used by using any other variable.

x = 20  

When we assign the price to the x. the integer object 10 is created in the Heap reminiscence and its reference is assigned to x.

x = 20  
y = x   
if id(x) == id(y):   
    print("The variables x and y are referring  to the same object")  

In the above code, we have assigned y = x, which potential the y object will refer to the equal object because Python allotted the equal object reference to new variable if the object is already exists with the equal value.

Now, see another example.

Example –

x = 20  
y = x  
x += 1  
If id(x) == id(y):  
      print("x and y do not refer to  the same object")  


x and y do not refer to the same object

The variables x and y are not referring the equal object due to the fact x is incremented by one, x creates the new reference object and y nonetheless referring to 10.

Transforming the Garbage Collector

The Python Garbage collector has labeled the objects using its generation. Python Garbage collector has the three-generation. When we outline the new object in the program, its existence cycle is dealt with by using the garbage collector’s first generation. If the object has use in a different program, it will be stimulated up to the subsequent generation. Every generation has a threshold.

The garbage collector comes into motion if the threshold of the number of allocations minus the wide variety of de-allocation is exceeded.

We can adjust the threshold cost manually the use of the GC module. This module provides the get_threshold() approach to check the threshold price of a distinct technology of the garbage collector. Let’s recognize the following example.

Example –

Import GC  


(700, 10, 10)

In the above output, the threshold cost 700 is for the first generation and different values for the 2nd and 0.33 generation.

The threshold fee for trigger the rubbish collector can be modified the use of the set_threshold() method.

Example – 2

import gc  
gc.set_threshold(800, 20, 20)  

In the above example, the cost of the threshold multiplied for all three generations. It will affect the frequency of going for walks the rubbish collector. Programmers don’t need to fear about the rubbish collector, but it plays imperative function in optimizing the Python runtime for the goal system.

Python rubbish collector handles the low-level details for the developer.

Importance of Performing Manual Garbage Collection

As we have discussed earlier, the Python interpreter handles the reference to object used in the program. It automatically frees the reminiscence when the reference remember will become zero. This is a classical method for reference counting, if it fails to work when the application has referenced cycles. The reference cycle happens when one or greater objects are referenced to every other. Hence the reference depend in no way will become zero.

Let’s understand the following example –

def cycle_create():  
    list1 = [18, 29, 15]  
    return list1  
[18, 29, 15, [...]]  

We have created the reference cycle. The list1 object is referring the object list1 itself. When the function returns object list1, the memory for the object list1 is not freed up. So that reference counting is no longer suitable for solving the reference cycle. But, we can resolve it by means of altering the garbage collector or overall performance of the rubbish collector.

To accomplish that, we will use the gc.collect() characteristic for the gc module.

import gc  
n = gc.collect()  
print("Number of object:", n)  

The above code will provide the wide variety of accrued and de-allocated objects.

We can function the manual rubbish collector the use of the two strategies – time-based or event-based garbage collection.

The gc.collect() method is used to perform the time-based garbage collection. This approach is referred to as after a constant time interval to perform time-based garbage collection.

In the even-based rubbish collection, the gc.collect() function calls after an match occur. Let’s apprehend the following example.

Example –

import sys, gc  
def cycle_create():  
    list1 = [18, 29, 15]  
def main():  
    print("Here we are creating garbage...")  
    for i in range(10):  
    print("Collecting the object...")  
    num = gc.collect()  
    print("Number of unreachable objects collected by GC:", num)  
    print("Uncollectable garbage:", gc.garbage)  
if __name__ == "__main__":  


Here, we are creating garbage... 
Collecting the object... 
Number of unreachable objects collected by GC: 10 
Uncollectable garbage: []

In the above code, we have created the list1 object referred through the list variable. The first element of the list object refers to itself. The list object’s reference be counted is always higher than zero, even it is deleted or out of scope in the program.

C Python Memory Management

In this section, we will talk about the C Python reminiscence structure in detail.

As we mentioned earlier, there is a layer of abstraction from the bodily hardware to Python. Various application or Python get right of entry to the virtual reminiscence that is created by the working system.

Python uses a element of the memory for internal use and non-object memory. Another phase of the reminiscence is used for Python object such as int, dict, list, etc.

CPython incorporates the object allocator that allocates reminiscence inside the object area. The object allocator gets a call each and every time the new object needs space. The allocator principal designs for small amount of statistics due to the fact Python does not contain too a lot records at a time. It allocates the memory when it is honestly required.

There are three important factors of the CPython memory allocation strategy.

Arena – It is the largest chunks of reminiscence and aligned on a web page boundary in memory. The working system makes use of the web page boundary which is the side of a fixed-length contiguous chuck of memory. Python assumes the system’s web page dimension is 256 kilobytes.

Pools – It is composed of a single size class. A pool of the identical dimension manages a double-linked list. A pool ought to be – used, full, or empty. A used pool consists of reminiscence blocks for records to be stored. A full pool has all the allotted and incorporate data. An empty pool does not have any statistics and can be assigned any dimension category for the block when needed.

Blocks – Pools carries a pointer to their “free” block of the memory. In the pool, there is a pointer, which suggests the free block of memory. The allocator doesn’t contact these block till it is without a doubt needed.

Common Ways to Reduce the Space Complexity

We can comply with some quality practices to minimize the space complexity. These techniques are supposed to shop quite space and make the application efficient. Below are a few practices in Python for memory allocators.

Avoid List Slicing

We define a listing in Python; the memory allocator allocates the Heap’s memory according to the listing indexing, respectively. Suppose we want a sub-list to the given list, then we perform the listing slicing. It is a simple way to get the sublist from the unique list. Somehow, it is appropriate for the small amount of facts but now not for the massive data.

Hence, listing slicing generates the copies of the object in the list. It simply copies the reference to them. As a result, the Python reminiscence allocator creates a copy of the object and allocates it. So we want to keep away from the list slicing.

The nice way to avoid the developer have to attempt to use the separate variable to song indices rather of cutting a list.

Use List Indexing Carefully

The developer ought to attempt to use the “for item in array” rather of “for index in range(len(array))” to save space and time. If our application doesn’t need the indexing of the list element, then don’t use it.

String Concatenation

String concatenation is now not suitable for saving house and time complexity. When possible, we ought to keep away from using ‘+’ for the string concatenation because strings are immutable. When we add the new string to the present string, Python creates the new string and allocates it to a new address.

Each string needs a constant measurement of reminiscence based on the personality and its length. When we alternate the string, it desires a one-of-a-kind amount of memory and requires reallocating.

Let’s run the following example.

a = Mango  
a = a + " Ice-cream"  
print (a)  


Mango Ice-cream

It will create variable a to refer to the string object, which is the string cost information.

Then we add the new sting in it the use of the ‘+’ operator. Python reallocates the new string in the memory based on its measurement and length. Suppose the unique string’s memory dimension is n byte, then the new string will be the m bytes.

Instead of using the string concatenation, we can use the” .join(iterable_object)” or structure or %. This makes a big influence on saving memory and time.

Use Iterators and Generators

Iterators are very helpful for both time and reminiscence when working on a large set of data. Working with the massive dataset, we want data processing right now and can wait for the application to technique the complete records set first.

Generators are the distinct features used to create the iterator function.

In the following example, we put into effect an iterator that calls the one of a kind generator function. The yield key-word returns the contemporary value, moving to the subsequent price solely on the loop’s next iteration.

Example –

def __iter__(self):  
    ''' This function allows are set to be iterable. Element can be looped over using the for loop'''  
    return self. _generator()  
def _generator(self):  
    """ This function is used to implement the iterable. It stores the data we are currently on and gives the                      next item at each iteration of the loop."""  
    for i in self.items():  
        yield i  

Use the Built-in Library when Possible

If we use techniques that have already been predefined in the Python library, then import the corresponding library. It would keep a lot of area and time. We can also create a module to outline the characteristic and import it to the current working program.


In this tutorial, we have discussed working of the reminiscence internally in Python. We have learned how Python manages reminiscence and additionally discussed default Python implementation. CPython is written in the C programming language. Python is a dynamic kind of language that used the Heap data structure to keep the memory.