Random() efficiency in C++

Question

I am writing function where I need to find the random number between 1 - 10. One of the easiest way is to use random() libc call. I am going to use this function a lot. But I don't know how efficient it will be. If any one has idea about efficiency of random() that will be a help ?

Also I notice that random() give the same pattern in 2 runs.

int main()
{
   for(int i=0;i<10;i++)
   {
    cout << random() % 10 << endl;     
   }
}

Output 1st time :- 3 6 7 5 3 5 6 2 9 1

Second time also I got same output.

Then how come it's random ?

Use a profiler if you're worried about performace. If you run the randomizer with the same seed, it will always give the same output. — Almo, Jan 31 '15 at 18:53
Actually I used Linux perf but there it's is not showing anything (like where it's consuming time). It's just showing how much CPU cycle it consume and it was around 1.73% of total CPU. — eswaat, Jan 31 '15 at 18:57
"efficiency" as in "performance" or "efficiency" as in "how good are the random numbers"? The latter is clearly a case of "you need to use srand()" [or similar] to start the random number sequence at a different point. "Good" random numbers is not trivial, so you may need further work if you need truly good random numbers [e.g. games where predictability becomes an issue, or using the numbers for scientific purposes where the distribution is important] — Mats Petersson, Jan 31 '15 at 19:01

score 6 · Accepted Answer · edited Jan 31 '15 at 22:18

Others have explained why it's the same sequence every time, but this is how you generate a random number with C++:

#include <random>

int main() {
    std::random_device rd{}; //(hopefully) truly random device
    std::mt19937 engine{rd()}; //seed a pseudo rng with random_device
    std::uniform_int_distribution<int> d(1,10); //1 to 10, inclusive
    int RandNum = d(engine); //generate
    return 0;
}

http://en.cppreference.com/w/cpp/numeric/random

score 2 · Answer 2 · edited May 23 '17 at 11:43

2

The actual execution time depends on your platform of course, but it is pretty much straight forward, couple multiplication and divisions or shifts:

What common algorithms are used for C's rand()?

I don't think you should be worried. If you need a lot of random numbers, then another random source probably would be a better choice for you.

If you are looking for tweaks, how about splitting the result from rand() into individual digits to get several results per call.

edited May 23 '17 at 11:43

Community

1
1

answered Jan 31 '15 at 19:51

Christopher Oezbek

23,994
6
61
85

The idea of splitting the result from `rand()` into individual digits (say, blocks of 4 bits to get random numbers from 1-10, say, by ignoring 0000, 1111, 1110, 1101, 1100 and 1011) is an interesting one but it would be important to [test the quality of the random numbers](http://en.wikipedia.org/wiki/TestU01) generated this way because the lower bits in most pseudo-random numbers are typically less random than the higher-valued bits. – Simon Jan 31 '15 at 20:25

score 0 · Answer 3 · answered Jan 31 '15 at 22:15

This way is very simple and effective, you only need to set the seed:

#include <iostream>
#include <stdlib.h>
#include <time.h>

using namespace std;

int main(){
    srand(time(NULL));
    for(int i=0;i<10;i++)
        cout << rand() % 10 << endl;     
}

score 0 · Answer 4 · answered Feb 01 '15 at 06:48

0

To fix the problem of getting same pattern in 2 runs just add the function randomize()

answered Feb 01 '15 at 06:48

Arun A S

6,421
4
29
43

Random() efficiency in C++

4 Answers4