2
\$\begingroup\$

I have written a program in Python 3 for the INTEST problem on Codechef (included in question). The program is taking a lot of time (55.76s) in Python 3. The same code takes almost 1/10 of the time time (4.97s) in C.

The purpose of this problem is to verify whether the method you are using to read input data is sufficiently fast to handle problems branded with the enormous Input/Output warning. You are expected to be able to process at least 2.5MB of input data per second at runtime.

Input

The input begins with two positive integers \$n\$ \$k\$ (\$n, k\le10^7\$). The next \$n\$ lines of input contain one positive integer \$t_i\$, not greater than \$10^9\$, each.

Output

Write a single integer to output, denoting how many integers \$t_i\$ are divisible by \$k\$.

Example

Input:

7 3
1
51
966369
7
9
999996
11

Output:

4

My questions are:

  1. How is it that the Python 3 code is taking so much time or which line of code specifically is taking the most time?
  2. Any modifications that can speed up the code?

My CodeChef submissions with time and code.

import sys
__author__ = 'Gourav Chawla'
"""
    Problem Code: INTEST
    Problem URL: http://www.codechef.com/problems/INTEST
    Compatability: Python 3.x
"""

n, k = input().split()
n = eval(n)
k = eval(k)

inputVar = 0
count = 0

# inputVar = [eval(x) for x in input().split()]

inputVar = list(map(int, sys.stdin.readlines()))

for i in inputVar:
   if i % k == 0:
       count += 1

print(count)
\$\endgroup\$
4
  • \$\begingroup\$ Welcome to Code Review! You could add the description of the problem in your question. \$\endgroup\$
    – Marc-Andre
    Jun 11, 2015 at 15:07
  • \$\begingroup\$ Oh, okay. But isn't the description too long for the question? \$\endgroup\$ Jun 11, 2015 at 15:16
  • \$\begingroup\$ See this question on meta for more information : meta.codereview.stackexchange.com/questions/1993/… . No it's not too long, it's important to understand your code to have the definition of the problem. \$\endgroup\$
    – Marc-Andre
    Jun 11, 2015 at 15:18
  • 1
    \$\begingroup\$ Got your point. Made the necessary changes. \$\endgroup\$ Jun 11, 2015 at 15:59

1 Answer 1

1
\$\begingroup\$

I think that there must be something wrong with the way CodeChef runs Python, or with the way it reports the times. The time limit for the problem is 8 seconds, but your Python submissions are passing even though they take 55 seconds and so ought to be out of time even with the five times allowance given to Python programs.

On my laptop, if I create about 80 megabytes of input:

import random
n, k = 10**7, 17
f = open('cr93327.data', 'w')
f.write('{} {}\n', n, k)
for _ in range(n):
    f.write('{}\n'.format(random.randrange(0, 10**7)))

Then this C program runs in 3.3 seconds:

#include <stdio.h>

int main() {
    unsigned n, k, i, m, count = 0;
    scanf("%u %u", &n, &k);
    for (i = 0; i < n; i++) {
        scanf("%u", &m);
        count += (m % k == 0);
    }
    printf("%u\n", count);
    return 0;
}

And this Python 3 program runs in 6.1 seconds:

import sys

n, k = map(int, next(sys.stdin.buffer).split())
count = 0
for i in map(int, sys.stdin.buffer):
    if not i % k:
        count += 1
print(count)

So Python is about twice as slow as C, not ten times as slow.

\$\endgroup\$
1
  • \$\begingroup\$ The codechef problem thing seems to be right because when I tried the same code on Spoj, it gave me 1.9s time. But then again Spoj might be using different servers and processing speed may differ on the sites in question hence the different output. \$\endgroup\$ Jun 15, 2015 at 0:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.