performance-testing（性能测试）

Performance testing

Running benchmarks

cd tools/performance

# Run all benchmarks (~15-20 minutes)
dotnet run -c Release

# Run specific category
dotnet run -c Release --filter *StreamIndicators*
dotnet run -c Release --filter *BufferIndicators*
dotnet run -c Release --filter *SeriesIndicators*

# Run specific indicator
dotnet run -c Release --filter *.EmaHub

Adding benchmarks

Series pattern

[Benchmark]
public void ToMyIndicator() => quotes.ToMyIndicator(14);

Stream pattern

[Benchmark]
public object MyIndicatorHub() => quoteHub.ToMyIndicatorHub(14).Results;

Buffer pattern

[Benchmark]
public MyIndicatorList MyIndicatorList() => new(14) { quotes };

Style comparison

[Benchmark]
public IReadOnlyList<MyResult> MyIndicatorSeries() => quotes.ToMyIndicator(14);

[Benchmark]
public IReadOnlyList<MyResult> MyIndicatorBuffer() => quotes.ToMyIndicatorList(14);

[Benchmark]
public IReadOnlyList<MyResult> MyIndicatorStream() => quoteHub.ToMyIndicator(14).Results;

Performance targets

Note: These are optimization goals for future v3.1+ effort. Current implementations vary—see PERFORMANCE_ANALYSIS.md for actual measured performance. Some indicator families (e.g., EMA) have inherent framework overhead due to simple operation costs.

Expected execution times (502 periods)

Note: These are optimization targets. Actual execution times vary by indicator complexity and current implementation.

Regression detection

# Auto-detect baseline and results
pwsh detect-regressions.ps1

# Custom threshold (default 10%)
pwsh detect-regressions.ps1 -ThresholdPercent 15

Exit codes:

• 0 - No regressions
• 1 - Regressions found

Creating baselines

cp BenchmarkDotNet.Artifacts/results/Performance.*-report-full.json \
   baselines/baseline-v3.0.0.json

Required optimization patterns

• Minimize allocations in hot paths
• Avoid LINQ in performance-critical loops
• Use Span<T> for zero-copy operations
• Cache calculations when possible
• Test with realistic data sizes (502 periods)

Prohibited patterns

• Excessive LINQ in hot paths
• Boxing/unboxing of value types
• Unnecessary string allocations
• Redundant calculations in loops
• Poor cache locality

See references/benchmark-patterns.md for detailed patterns.