c# - Why loop on array object with `foreach` is faster than lambda `ForEach`?

Question

I work on an array that and I have to loop over it. First, I use lambda ForEach

Array
.ForEach<int>( array, ( int counter ) => {
    Console.WriteLine( counter ); 
} );

and then I use simple foreach. I find that simple foreach is so faster than lambda ForEach, but when i test it with generic lists, ForEach is faster than simple foreach.

why loop on array object with foreach is faster than lambda ForEach? Update: I test on array

Answer 1

I edited Keith's code a bit - On my machine foreach performed about six times faster than Array.ForEach:

class Program
{
    static void Main(string[] args)
    {
        Benchmark(50);
    }

    private static void Benchmark(int iterations)
    {
        int[] list = Enumerable.Range(0, 100000000).ToArray();

        long sum = 0;
        for (int i = 0; i < iterations; i++)
        {
            sum += ArrayForeach(list);
        }

        Console.WriteLine("ForEach " + sum / iterations);

        sum = 0;
        for (int i = 0; i < iterations; i++)
        {
            sum += Foreach(list);
        }

        Console.WriteLine("foreach " + sum / iterations);
    }

    private static long Foreach(int[] list)
    {
        long total = 0;
        var stopWatch = Stopwatch.StartNew();
        foreach (var i in list)
        {
            total += i;
        }
        stopWatch.Stop();
        return stopWatch.ElapsedTicks;
    }

    private static long ArrayForeach(int[] list)
    {
        long total = 0;
        var stopWatch = Stopwatch.StartNew();
        Array.ForEach(list, x => total += x);
        stopWatch.Stop();
        return stopWatch.ElapsedTicks;
    }
}

On my machine (which might run a different CLR than others) it produces (in Release):

ForEach 695910  
foreach 123852  

In Debug:

ForEach 941030
foreach 845443

• It shows that foreach enjoys some compiler optimizations, I guess mainly in accessing the list in memory.
• In Debug, it looks like the overhead of running the lambda, and passing the number (by value) is responsible for the difference.

I suggest that someone with more time take a look with Reflector...

Answer 2

I find the lambda faster in my testing. Copy pasting MSDNs stopwatch code and decorating it with two versions of iterating a List.... ( I also changed the order of which test goes first, and I get the same timings). The Linq based iteration with a lambda goes faster.

Lambda  00:00:00.49
foreach 00:00:00.58

and the code..

var list = Enumerable.Range(0, 100000000).ToArray();
        var total = 0;
        var stopWatch = new Stopwatch();
        stopWatch.Start();
        Array.ForEach(list, x => total += x);

        stopWatch.Stop();
        TimeSpan ts = stopWatch.Elapsed;
        string elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:00}",
            ts.Hours, ts.Minutes, ts.Seconds,
            ts.Milliseconds / 10);
        Console.WriteLine("RunTime " + elapsedTime);


        stopWatch = new Stopwatch();
        stopWatch.Start();
        foreach (var i in list)
        {
            total += i;
        }      
        stopWatch.Stop();
        // Get the elapsed time as a TimeSpan value.
        ts = stopWatch.Elapsed;

        // Format and display the TimeSpan value. 
        elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:00}",
            ts.Hours, ts.Minutes, ts.Seconds,
            ts.Milliseconds / 10);
        Console.WriteLine("RunTime " + elapsedTime);