Back to List

Have Your Pi and Eat it Too: A Comparison of C# and C++ Performance

John Ptacek John Ptacek  |  
Mar 14, 2013
 

​In honor of National Pi Day, we thought it would be fun to calculate Pi and talk about performance differences between C# and C++. There is some math coming your way (yeah, MATH!), so feel free to skip to the end for a discussion on performance if math isn’t your thing.

So, back in the day, when things where WAY different, Pi was calculated using a mathematical series. One of the more famous ones uses the arctan math function to calculate the value of π. This is called the Leibniz formula for π and is represented by:

Now, this is one of the more inefficient ways to calculate π, it takes about 10 million terms to get an accuracy of 7 decimal places. For those of us asking for performance differences between languages though, that is great!

We have written two programs to calculate π, one in C# and one in C++. They are both syntactically similar and use the same algorithm. We tried to get as close to an apples to apples comparison as we could, but in comparing two different languages, while we did our best, you are still comparing an apple to an orange.

So, what did we see? We calculated π to 15 decimal points, 3.141592653589793 for those of you keeping track at home. Both programs took almost 100 million iterations to converge … 99,995,330 to be exact. We ran each of the programs 10 times to get an average as shown in the following table:

The calculation of π using C++ is 35% faster that it’s corresponding C# code. This is to be expected. Many modern languages such as C# and Java run through a managed runtime. This brings MANY efficiencies to developers. However, the runtime tends to add overhead when it comes to performance.

The pragmatic folks reading this realize that the benefits of a managed execution environment and modern language generally outweigh performance gains from a language like C++. There are reasons that C# and Java are used instead of C++ because developers can be more productive. Our example here, using a very slow numerical algorithm and running it for ten million iterations is a stretch for most scenarios. The 2.5 second difference for such an intensive calculation, while relatively large in our example, in the grand scheme of things is not that big of a difference.

So why bring it up? It is important to think through performance more and more in today’s computing environment for two reasons. The first is cost. As organizations move to utility-based computing with hosted services, such as Amazon Web Services (AWS) and Microsoft Azure, they are paying for their compute cost. If you can realize performance gains that reduce CPU cycles it has a direct impact on the bottom line. The second reason is that mobile is taking over the world, and mobile users want more battery life, not less. Fewer CPU cycles on a smartphone or tablet will result in better battery life, which makes everyone happy.

It will be very rare that most organizations would use C++ or other “closer to the metal” languages for their Line of Business (LOB) applications. The moral of the story though is for intensive processes in cloud-hosted or mobile environments, you want to be razor focused on optimizing your code for performance, no matter what language you choose.

Both code examples are included below, so for National Pi day, make a circle, calculate some π, eat some pie and enjoy!

C++

// LifeOfPi.cpp : Defines the entry point for the console application.
//
 
#include "stdafx.h"
#include <iostream>
#include <iomanip>
using namespace std;
 
void pi( double *pi_value, int *iterations); 
 
int main ( void ) 
 
      int iterations = 0;     // number of actual loop iterations in pi
      double pi_value = 0;    // computed value of pi
 
      clock_t cBegin = (double)clock () ;
      pi(&pi_value, &iterations); 
      printf("   %0.15f with %d iterations\n",  pi_value, iterations); 
      clock_t cEnd = (double) clock() ;
 
      std::cout << float( cEnd - cBegin )
 
 
return 0; 
 
}   
 
void pi( double *pi_value, int *iterations) 
      int i; 
      int k=1;
      double realPi = 3.141592653589793;
      double epsilon = 0.00000001;
      double delta=10;
      *pi_value= 0;
      while (delta>epsilon) {
            *pi_value += pow(-1,(k+1))/(2*k-1);
            delta = fabs((4* *pi_value)-realPi);
            k++;
      }
      *iterations = k;
      *pi_value = 4 * *pi_value;
 
 
      return; // indicate program ended sucessfully
 
} // end fucntion main
C#
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
 
namespace LifeOfPiCSharp
{
    class Program
    {
        static void Main(string[] args)
        {
            int iterations = 0;
            double pi_value = 0;
 
            DateTime startTime = DateTime.Now;
            pi(out pi_value,out iterations);
            DateTime endTime = DateTime.Now;
            Console.WriteLine(pi_value.ToString() + " in " + iterations);
            TimeSpan diff = endTime.Subtract(startTime);
            Console.WriteLine(diff.Seconds + "."+diff.Milliseconds);
        }
 
        public static void pi(out double pi_value, out int totalIterations)
        {
            double realPi = 3.141592653589793;
            double epsilon = 0.00000001;
            double delta = 10;
            int k = 1;
           
            pi_value = 0;
            while (delta > epsilon)
            {
                pi_value += Math.Pow(-1, (k + 1)) / (2 * k - 1);
                delta = Math.Abs((4 * pi_value) - realPi);
                k++;
            }
            totalIterations = k;
            pi_value = 4 * pi_value;
 
        }
    }
}



 

Love our Blogs?

Sign up to get notified of new Skyline posts.

 


Related Content


Spring 2019 Kentico User Group
Apr 17, 2019
Location: Waukesha County Technical College - Pewaukee Campus - 800 Main Street, Pewaukee, Wisconsin 53072 - Building: Q, Room: Q361
Blog Article
How to Build a Custom Android ListAdapter
Michael FazioMichael Fazio  |  
Mar 30, 2021
Michael Fazio is a tech speaker and senior software developer at Skyline Technologies - a Core BTS Company. He’s been the Android lead for multiple billion-dollar companies and has seen the Android development process evolve from far too many Activities and a nigh unusable emulator to...
Blog Article
Why I Stopped Testing My Code (and You Should Too)
Todd TaylorTodd Taylor  |  
Feb 23, 2021
About the author: Todd M. Taylor is a Senior Software Engineer with a passion for producing quality software using Microsoft Azure technologies.   I confess that the title of this blog post is clickbait. I feel a little ashamed that I've stooped so low to get your attention.But as...
Blog Article
Developer Playbook for Getting Started with DevOps
Bob SchommerBob Schommer  |  
Jan 26, 2021
About the author: Bob Schommer is a Senior Consultant with over 10 years of agile experience who has trained over 300 professionals on agile and DevOps practices. He has coached and advised large and small organizations to optimize their software delivery chain. The buzzword to end all buzzwords...
Blog Article
Manager Playbook for Getting Started with DevOps
Bob SchommerBob Schommer  |  
Jan 12, 2021
About the author: Bob Schommer is a Senior Consultant with over 10 years of agile experience who has trained over 300 professionals on agile and DevOps practices. He has coached and advised large and small organizations to optimize their software delivery chain. Agile practices and DevOps...