DotNet.Benchmarks

High-performance benchmarks for .NET features using BenchmarkDotNet.

Projects

Project	Description	Target
DotNet.Benchmarks.Compression	GZip vs Brotli vs Zstandard (native .NET 11)	net11.0
DotNet.Benchmarks.GuidGen	GUID v4 vs GUID v7 (generation, index locality, SQLite, hotspots)	net11.0

Structure

DotNet.Benchmarks/
├── src/
│   ├── DotNet.Benchmarks.Compression/
│   │   ├── Benchmarks/
│   │   │   └── ZstdVsBrotliBench.cs
│   │   ├── Program.cs
│   │   └── DotNet.Benchmarks.Compression.csproj
│   └── DotNet.Benchmarks.GuidGen/
│       ├── Benchmarks/
│       ├── Scenarios/
│       ├── Program.cs
│       └── DotNet.Benchmarks.GuidGen.csproj
├── results/
│   ├── Compression/
│   │   ├── artifacts/
│   │   ├── BenchmarkDashboard.html
│   │   └── CompressionResults.md
│   └── GuidGen/
│       ├── artifacts/
│       └── GuidScenarioReport.md
├── docs/
├── DotNet.Benchmarks.slnx
└── README.md

Benchmark Results: .NET 11 Compression

Environment:

• OS: Linux (Ubuntu 24.04)
• Framework: .NET 11.0 Preview
• Hardware: Cloud VPS (12 vCPUs, 48 GB RAM, SSD)

Summary

• Zstandard (Zstd) is significantly faster than GZip and Brotli in Optimal mode (up to 4x faster).
• Zstandard also wins in Decompression speed across all sizes.
• Brotli provides good compression but is much slower to compress in Optimal mode.

Highlights (100 MB Payload)

Algorithm	Mode	Time (ms)	Throughput
Zstandard	Optimal	591 ms	~169 MB/s
GZip	Optimal	1073 ms	~93 MB/s
Brotli	Optimal	2549 ms	~39 MB/s

Lower time is better.

You can view the full interactive charts in results/BenchmarkDashboard.html.

Prerequisites

• .NET 11 Preview SDK
  • The project auto-resolves SDKs via global.json.
  • If using a user-local installation, ensure DOTNET_ROOT points to it:

    $env:DOTNET_ROOT = "$env:USERPROFILE\.dotnet"
    $env:PATH = "$env:USERPROFILE\.dotnet;$env:PATH"

How to Run

Important: Benchmarks must always be run in Release mode for reliable results.

Verify Setup

dotnet --version
# Expected output: 11.0.100-preview.1... or similar

dotnet --list-sdks
# Must show 11.0.100-preview.1.26104.118

---

Compression Benchmark

Full run (statistical analysis with multiple payload sizes):

dotnet run --project src/DotNet.Benchmarks.Compression -c Release

The results will appear in the console with columns for Mean (average time), Error, StdDev, Ratio, and Allocated (memory).

Artifacts path: results/Compression/artifacts/

Quick mode (smoke test, shorter execution):

dotnet run --project src/DotNet.Benchmarks.Compression -c Release -- --job short

---

GUID Benchmark (v4 vs v7)

Option A: Full BenchmarkDotNet run (comprehensive, slower)

dotnet run --project src/DotNet.Benchmarks.GuidGen -c Release

Outputs detailed benchmarks with statistical analysis. Artifacts: results/GuidGen/artifacts/

Option B: Scenario Report (quick consolidated table, recommended for demos)

dotnet run --project src/DotNet.Benchmarks.GuidGen -c Release -- --scenarios

This generates a practical table comparing v4 vs v7 across 4 scenarios (generation, locality, SQLite insert, hotspot risk).

Output:

• BenchmarkDotNet artifacts: results/GuidGen/artifacts/
• Scenario report: results/GuidGen/GuidScenarioReport.md

Example Session

# Clone and enter repo
cd DotNet.Benchmarks

# Verify SDK
dotnet --version

# Run compression (quick)
dotnet run --project src/DotNet.Benchmarks.Compression -c Release -- --job short
# Check results/Compression/artifacts/ for HTML dashboard

# Run GUID scenarios (fast, informative)
dotnet run --project src/DotNet.Benchmarks.GuidGen -c Release -- --scenarios
# Check results/GuidGen/GuidScenarioReport.md for consolidated table

# Run full GUID benchmarks (comprehensive, slower ~1-2 min)
dotnet run --project src/DotNet.Benchmarks.GuidGen -c Release
# Check results/GuidGen/artifacts/ for statistical analysis

License

See LICENSE.txt.