Let's talk about microbenchmarking

This talk is about a wonderful world of microbenchmarking and microperformance. Andrey will discuss why microbenchmarking is hard and how to measure very quick operations right way. Also, he will talk about the BenchmarkDotNet library which can help to design correct microbenchmarks. Content: General methodology; Common pitfalls and mistakes; How long we should measure and how many times?; Warmup importance; Why we should run benchmarks in different environments (different C# Compilers/CLR versions/JITs/Operation systems/etc); Overhead evaluation; Statistics calculation; Good practices.

Download presentation.

Andrey Akinshin andrey_akinshin

JetBrains

Andrey is a .NET MVP, a silver medalist of ACM ICPC, the maintainer of BenchmarkDotNet, a fan of micro-optimizations and one of the developers of Rider at JetBrains.