Python numba 的使用

• 挺好，原作者太忙了，作为他的... 我来代替更一下。
• 祝他半期不挂！

• 谁曾侍秋风画扇。遮掩几时落寞。短梳牛角理红妆。阶前梧桐老，扉下虫丝长。
• 点墨胸中话短长。南北天桥谁架？再相逢羞谈过往。今朝晨露好，融化笺上霜。

cProfile测试

• 很好，这是一段测试代码

import cProfile
from time import sleep
 
def delay1():
    sleep(.5)
    
def delay2():
    sleep(1)
    
def simulate(Input):
 
    while_count = 10000000
    while while_count:
        while_count -= 1
        Input += 1
        
    for_count = 10000000
    for i in range(for_count):
        Input += 2
 
    for i in range(for_count):
        Input += 2
 
 
    delay1()
    delay2()
    delay2()
    delay2()
    delay2()
    
    return Input

---

• 测试命令

>>>cProfile.run('simulate(21)')

• 输出

         14 function calls in 5.770 seconds
 
   Ordered by: standard name
 
   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    1.241    1.241    5.770    5.770 0.py:11(simulate)
        1    0.000    0.000    0.503    0.503 0.py:5(delay1)
        4    0.000    0.000    4.026    1.006 0.py:8(delay2)
        1    0.000    0.000    5.770    5.770 <string>:1(<module>)
        1    0.000    0.000    5.770    5.770 {built-in method builtins.exec}
        5    4.529    0.906    4.529    0.906 {built-in method time.sleep}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}

• ncall：函数运行次数
• tottime： 函数的总的运行时间减去函数中调用子函数的运行时间 
• 第一个percall：percall = tottime / nclall
• cumtime:函数及其所有子函数调整的运行时间
• 第二个percall：percall = cumtime / nclall 

numba的使用

• 模拟函数

import cProfile
import time
from numba import jit
 
@jit
def simulate(Input):
    while_count = 10000000
    while while_count:
        while_count -= 1
        Input += 1
    for_count = 10000000
    for i in range(for_count):
        Input += 2
    for i in range(for_count):
        Input += 2
    return Input
 
start = time.time()
simulate(100)
end = time.time()
print(end-start)

• 时间

0.31107425689697266
cProfile.run('simulate(100)')
         4 function calls in 0.000 seconds
 
   Ordered by: standard name
 
   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    0.000    0.000 0.py:5(simulate)
        1    0.000    0.000    0.000    0.000 <string>:1(<module>)
        1    0.000    0.000    0.000    0.000 {built-in method builtins.exec}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}

• 未优化前

1.029198408126831
cProfile.run('simulate(100)')
         4 function calls in 1.234 seconds
 
   Ordered by: standard name
 
   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    1.234    1.234    1.234    1.234 0.py:6(simulate)
        1    0.000    0.000    1.234    1.234 <string>:1(<module>)
        1    0.000    0.000    1.234    1.234 {built-in method builtins.exec}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}

• @njit和@jit(nopython=True)具有相类似的功能

numba和numpy的结构体