Experiment environment

OS	macOS Sierra
CPU	2.3 GHz Intel Core i7
memory	16 GB 1600 MHz DDR3
Python version	2.7.10

Experiment

Preparation

Prepare the following slightly larger data and functions, and check the execution time, object identifier after calculation, and bytecode, respectively.
Call time.time () before and after the function call to measure the execution time by the difference
Use the built-in function ʻid () `for the object identifier after the operation.
Use the dis module to check the bytecode

def add(list_x, list_y):                                                        
    list_x = list_x + list_y                                                    
    return list_x   

def inplace(list_x, list_y):                                                    
    list_x += list_y                                                            
    return list_x 

list_a = [[0 for x in range(10000)] for y in range(10000)]
list_b = [[0 for x in range(10000)] for y in range(10000)]

Experimental result

I made a code that outputs the following and confirmed it.

add(list_a, list_b)
Execution time: 0.000458
id(list_a): 4391953400 -> 5530614960
------ dis.dis(add) ------
  8           0 LOAD_FAST                0 (list_a)
              3 LOAD_FAST                1 (list_b)
              6 BINARY_ADD          
              7 STORE_FAST               0 (list_a)

  9          10 LOAD_FAST                0 (list_a)
             13 RETURN_VALUE
   
  
inplace(list_a, list_b)
Execution time: 0.000285
id(list_a): 5530614960 -> 5530614960
------ dis.dis(inplace)  ------
 44           0 LOAD_FAST                0 (list_a)
              3 LOAD_FAST                1 (list_b)
              6 INPLACE_ADD         
              7 STORE_FAST               0 (list_a)

 45          10 LOAD_FAST                0 (list_a)
             13 RETURN_VALUE

	add	inplace	result
Execution time	0.000458	0.000285	in place is faster
Bytecode difference	BINARY_ADD	INPLACE_ADD	You can see that it has changed even with bytecode
id(list_a)	There is a change	No change	add is created as a new object and replace is assigned a value to the same object.

Apparently add is slower because it has to allocate memory for new objects.
The difference is small, but considering the internal operation, I thought it would be better to use A + = B from the viewpoint of memory and execution speed.

motivation

I had never been aware of the internal movements of both parties, so I investigated it myself. As long as I'm embarrassed when I was thinking, "Let's do it together!"

Recommended Posts

I tried to find out the difference between A + = B and A = A + B in Python, so make a note

[Python] I tried to implement stable sorting, so make a note

I tried to find out the difference between A + = B and A = A + B in Python, so make a note

Let's make a simple game with Python 3 and iPhone

Build a Python environment and transfer data to the server

I tried to enumerate the differences between java and python

I tried to make a stopwatch using tkinter in python

I also tried to imitate the function monad and State monad with a generator in Python

[Introduction to Python] What is the difference between a list and a tuple?

About the difference between "==" and "is" in python

A note I looked up to make a command line tool in Python