BenchmarkDotNet-Tutorial.md

Background Info

A .NET library to make Benchmarking easy! See https://benchmarkdotnet.org/ and in-particular the Getting Started Guide the Main Features List

---

Hello World

Add the NuGet package Install-Package BenchmarkDotNet

Run the following code:

public class Program
{
    [Benchmark]
    public void Slow() => Thread.Sleep(100);

    [Benchmark]
    public void Fast() => Thread.Sleep(25);

    static void Main(string[] args)
    {
        Console.WriteLine("BenchmarkDotNet - Hello World!");
        var summary = BenchmarkRunner.Run<Program>();
    }
}

Challenges

• Using '+' on String (i.e. myString += "blah") compared to StringBuilder
  • Why is one faster that the other (Hint: see Diagnosers for some features that might help)
  • Does the size of the Strings involved make a difference (Hint: check out Parameterisation)
• Replicate benchmarks from Performance Improvements in .NET Core 2.1
  • Do you get the same results?
  • Can you compare .NET Core with .NET Framework (Hint: see Configs and Job.Core)
• Find a blog post explaining a .NET Performance 'Best Practice' i.e. like "use StringBuilder rather than concatenating Strings" (from above). Prove or Disprove the advice
  • Is it always true or does it depend on the version of .NET?
  • Does it depend on 32-bit or 64-bit? (Hint: Disassmbler will help)
  • Does the size of the inputs make a difference? (Hint: check out Parameterisation)
  • Bonus points for any graphs (Hint: see Exporters - Plots

---

Useful Links

• https://benchmarkdotnet.org/
• https://benchmarkdotnet.org/articles/samples/IntroBasic.html
• https://www.slideshare.net/mattwarren/where-the-wild-things-are-benchmarking-and-microoptimisations
• https://github.com/adamsitnik/awesome-dot-net-performance#benchmarking