| name | test-performance |
| description | Use when measuring or reviewing performance, benchmarks, load tests, stress tests, spike tests, performance regressions, or optimization claims. |
| last-updated | "2026-06-12T00:00:00.000Z" |
| difficulty | intermediate |
| tokens | ~0.5K |
Performance Testing
Performance testing with the repository's benchmark and load-test tooling.
When to Apply
- Benchmarking code performance.
- Load testing APIs.
- Measuring performance regressions.
- Validating optimization claims.
BenchmarkDotNet
Run benchmark commands only when the user explicitly allows build/test execution. Use the repository profile command when available; otherwise discover the benchmark project and set PERFORMANCE_PROJECT.
[MemoryDiagnoser]
public class MyBenchmark
{
private readonly List<int> _data = Enumerable.Range(0, 1000).ToList();
[Benchmark(Baseline = true)]
public int ForLoop()
{
var sum = 0;
for (var i = 0; i < _data.Count; i++) sum += _data[i];
return sum;
}
[Benchmark]
public int LinqSum() => _data.Sum();
}
dotnet run -c Release --project "$PERFORMANCE_PROJECT"
dotnet run -c Release --project "$PERFORMANCE_PROJECT" -- --filter-method Fully.Qualified.PerformanceTestClass.Run_Benchmarks
| Attribute | Purpose |
|---|
[Benchmark(Baseline = true)] | Baseline for comparison |
[MemoryDiagnoser] | Memory allocation metrics |
[GlobalSetup] / [IterationSetup] | Setup hooks |
[Params(1, 10, 100)] | Parameterized benchmarks |
K6 Load Testing
Run load-test commands only when the user explicitly allows load-test execution. Discover script paths from the repository profile or the k6/ folder.
k6 run k6/scripts/api-load-test.js
k6 run --vus 50 --duration 1m k6/scripts/api-load-test.js
k6 run --out json=results.json k6/scripts/api-load-test.js
| Test Type | Purpose | Config |
|---|
| Load | Normal load | vus: 10, duration: '5m' |
| Stress | Beyond capacity | Ramp up VUs until failure |
| Spike | Sudden load | Quick ramp to high VUs |
| Soak | Extended duration | Low VUs, long duration |
Best Practices
- Use release mode for micro-benchmarks.
- Avoid parallel benchmark execution unless the framework supports it.
- Document baselines to track regressions.
- Tie optimization claims to before/after measurements.
Related Skills
