Tensorflow的while循环比传统的while循环慢

下面是一个常规的while循环,它执行一个基本的add操作-

import time
def check(a,b):
 while(a<b):
    a += 1
 return [a,b]

a = 1
b = 1500000
start = time.time()
check(a,b)
print("Time = ",time.time() - start)

Time =  0.07060480117797852

下面是使用Tensorflow的优化代码-

import tensorflow as tf
def cond(t1, t2):
    return tf.less(t1, t2)

def body(t1, t2):
    return [tf.add(t1, 1), t2]

t1 = tf.constant(1)
t2 = tf.constant(1500000)

start = time.time()
res = tf.while_loop(cond, body, [t1, t2], parallel_iterations = 10)
print("Time = ",time.time() - start)

Time =  22.1693217754364