Unlocking Performance Improvements in .NET
summary
Shows concrete wins: microbenchmarks with BenchmarkDotNet prove Int32 formatting falls from ~48ns to ~11ns across releases; regex matching drops 470ns→148ns; compression gives 3–4x throughput and is backported to .NET Framework; TechEmpower fortunes jumps ~280k→400k requests/sec; ConcurrentQueue rewrite cuts 45ms→11ms and 8.5MB→hundreds of bytes; async/value-task work removes per-operation allocations (sockets go allocation-free); widespread adoption of Span<T> and hardware intrinsics halves or better common ops (e.g., string search ~41ns→6ns).
AI-written summary · about
chapters (12)
- 0:00Introduction to .NET Perf
- 3:40Performance blog posts
- 6:16Distribution impact
- 11:21Open source contribution
- 13:54Benchmarking methods
- 18:35Real-world optimizations
- 24:34Algorithms and rewrites
- 28:04Span and modernization
- 32:31Native to managed code
- 35:51GC and latency improvements
- 38:55Async and memory pooling
- 44:11Vectorization and conclusion
related talks
46mHigh Performance .NET
01h 03mTurbocharged: Writing High-Performance C# and .NET Code
37mRock Your Code Code and App Performance for Microsoft .Net
57mPragmatic Performance: When to care about perf, and what to do about it
59mPerformance tricks I learned from contributing to open source .NET packages
56mState of the .NET Performance
01h 00mRock Your Code: Code & App Performance for Microsoft .NET
01h 04mTurbocharged: Writing High-Performance C# and .NET Code