Amdahl's Law Example in C#

Question

I was working with some paralization and that brought me looking into Amdahl\'s law. I\'ve read a number of posts on the topic;

Calculate performance gains using Amdahl

Answer 1

Note: A complete working downloadable version of the program can be found on My Github Page

So with Amdahl's Law, we split the work in to "Work that must run in serial" and "Work that can be parallelized", so let's represent those two workloads as List:

var serialWorkLoad = new List { DoHeavyWork, DoHeavyWork };
var parallelizableWorkLoad = new List { DoHeavyWork, DoHeavyWork, DoHeavyWork, DoHeavyWork, DoHeavyWork, DoHeavyWork, DoHeavyWork, DoHeavyWork };

Where the DoHeavyWork delegate is abstracted brilliantly as:

static void DoHeavyWork()
{
    Thread.Sleep(500);
}

As you can see I've made the parallelizable workload a bit heavier for fun and to make a decent example of it.

Next we have to run both workloads in Serial to get our baseline:

var stopwatch = new Stopwatch();
stopwatch.Start();
// Run Serial-only batch of work
foreach (var serialWork in serialWorkLoad)
{
    serialWork();
}

var s1 = stopwatch.ElapsedMilliseconds;

// Run parallelizable batch of work in serial to get our baseline
foreach (var notParallelWork in parallelizableWorkLoad)
{
    notParallelWork();
}

stopwatch.Stop();
var s2 = stopwatch.ElapsedMilliseconds - s1;

At this point we have how long it took each workload to run in serial. Now, let's run it again, with the parallelizable portion parallelized.

stopwatch.Reset();
stopwatch.Start();
// Run Serial-only batch of work
foreach (var serialWork in serialWorkLoad)
{
    serialWork();
}

var p1 = stopwatch.ElapsedMilliseconds;

// Run parallelizable batch of work in with as many degrees of parallelism as we can
Parallel.ForEach(parallelizableWorkLoad, (workToDo) => workToDo()); // In Java this is Magic Unicorns

stopwatch.Stop();
var p2 = stopwatch.ElapsedMilliseconds - p1;

Now that we have the baseline and the parallelized version, we can calculate the speedup and report our findings:

var speedup = (double)(s1 + s2) / (p1 + p2);

Console.WriteLine("Serial took  : {2}ms, {0}ms for serial work and {1}ms for parallelizable work", s1, s2, s1 + s2);
Console.WriteLine("Parallel took: {2}ms, {0}ms for serial work and {1}ms for parallelizable work", p1, p2, p1 + p2);
Console.WriteLine("Speedup was {0:F}x", speedup);

And as Amdahl's Law tells you, it is hard to scale perfectly with the # of cores you have because of the serial-only work.