Jul 26, 2022

CPU profiling in python

1. using cProfile cli command

        python -m cProfile code.py
   

2. using cProfile.run method

        from time import sleep
   
        if __name__ == '__main__':
            import cProfile
            cProfile.run('sleep(1)', filename='result.out')
   

3. using pstat to see the result

        python -m pstat result.out
   

   Welcome to the profile statistics browser.
   code.out%

4. sort by total time

        sort tottime
   

5. top 3 position in statistic

        stats 3
   

   code.out% stats 3
   Wed Jul 27 08:47:41 2022 code.out

   15 function calls (14 primitive calls) in 1.003 seconds

   Ordered by: internal time
   List reduced from 12 to 3 due to restriction <3>
   ncalls tottime percall cumtime percall filename:lineno(function) 5
   1 1.002 1.002 1.002 1.002 {built-in method time.sleep}
   1 0.000 0.000 1.003 1.003 {built-in method builtins.exec}
   1 0.000 0.000 1.002 1.002 :1()

6. using pysnooper

   installation

        pip3 install pysnooper
   

   example

    import pysnooper
   
    @pysnooper.snoop()
    def test(number: int):
        sum = 0
        for i in range(number):
            sum += i
        return sum
   
    if __name__ == '__main__':
        test(1)