Performance Issues

First of all Microsoft Performance Explorer must not be used to measure execution times or performance comprasion.

I wanted to check the boxing and unboxing performance and tried to use Microsoft Performance Explorer result it sucks try it,you can’t trust its execution times.So i used this code

   1: public static class PerformanceTimer

   2: {

   3:     [DllImport("Kernel32.dll")]

   4:     public static extern void QueryPerformanceCounter(ref long ticks);

   5:  

   6:     static long startTime = 0;

   7:     static long stopTime = 0;

   8:  

   9:     public static void Start()

  10:     {

  11:         QueryPerformanceCounter(ref startTime);

  12:     }

  13:  

  14:     public static void Stop()

  15:     {

  16:         QueryPerformanceCounter(ref stopTime);

  17:     }

  18:  

  19:     public static long Time

  20:     {

  21:         get

  22:         {

  23:             return stopTime - startTime;

  24:         }

  25:     }

  26: }

and got real results,

1-Never use microsoft performance explorer

2-Don’t use boxing it is really 20 times slower

here are my execution times

1636417 ticks used boxing

82530 ticks did not use boxing

   1: class Program

   2: {

   3:  

   4:     static void Main(string[] args)

   5:     {

   6:         Stopwatch sw = new Stopwatch();

   7:         sw.Start();

   8:         PerformanceTimer.Start();

   9:         //ArrayList aList = new ArrayList(1000001);

  10:         //aList.Add(0);

  11:         //for (int i = 0; i < 999999; i++)

  12:         //{

  13:         //    aList.Add(i);

  14:         //    int a = Convert.ToInt32(aList[i]) + 2;

  15:         //    aList[i] = a;

  16:         //}

  17:         List<int> gList = new List<int>(999999);

  18:         for (int i = 0; i < 999999; i++)

  19:         {

  20:             gList.Add(i);

  21:             int a = gList[i] + 2;

  22:             gList[i] = a;

  23:         }

  24:         PerformanceTimer.Stop();

  25:         sw.Stop();

  26:         double ms = (sw.ElapsedTicks * 1000.0) / Stopwatch.Frequency;

  27:         Console.WriteLine(PerformanceTimer.Time.ToString());

  28:         Console.WriteLine(sw.ElapsedMilliseconds.ToString());

  29:         Console.WriteLine(ms.ToString());

  30:         Console.Read();

  31:     }

  32: }

used upper code to test performances, used arraylist as boxing example and used generic list as non boxing.

U can also use System.Diagnostics.Stopwatch;

3- Use Convert to it is better than implicit cast.

4-Use String builder instead of + to build large strings.

5-Use String.Equals instead of == and String.IsNullorEmpty instead of !=.

6-I tested for vs foreach i think there is no difference in execution times.(tested summing an array of 10000 random number previously generated).