/benchmark - Performance Benchmarking

Run performance benchmarks on code or endpoints.

Usage

/benchmark [target] [--type function|api|load]

Arguments:

• target - Function, file, or API endpoint to benchmark
• --type - Benchmark type (default: auto-detect)

Examples

/benchmark src/utils/parser.ts
/benchmark processData --type function
/benchmark /api/users --type api
/benchmark /api/search --type load

Benchmark Types

Function Benchmarks

• Execution time (avg, min, max, p95, p99)
• Memory allocation
• Garbage collection impact
• Comparison with baseline

API Benchmarks

• Response time
• Throughput (req/sec)
• Error rates
• Payload size analysis

Load Benchmarks

• Concurrent user simulation
• Sustained load testing
• Spike testing
• Resource utilization under load

Output Format

## Benchmark Results: processData()

**Target:** src/utils/parser.ts:processData
**Iterations:** 10,000
**Date:** 2024-01-15 10:30:00

### Execution Time
| Metric | Value |
|--------|-------|
| Average | 2.34ms |
| Median | 2.12ms |
| Min | 1.89ms |
| Max | 8.45ms |
| P95 | 3.21ms |
| P99 | 5.67ms |
| Std Dev | 0.89ms |

### Memory
| Metric | Value |
|--------|-------|
| Avg Allocation | 1.2MB |
| Peak | 4.8MB |
| GC Pauses | 12 |

### Throughput
- **Operations/sec:** 427
- **Data processed:** 42.7 MB/sec

### Comparison
| Version | Avg Time | Change |
|---------|----------|--------|
| Current | 2.34ms | - |
| Previous | 2.89ms | -19% faster |
| Baseline | 3.45ms | -32% faster |

### Visualization

Response Time Distribution: 1-2ms |████████████████████ 45% 2-3ms |██████████████ 32% 3-4ms |██████ 15% 4-5ms |██ 5% 5ms+ |█ 3%

### Recommendations
1. P99 latency is 2.4x average - investigate outliers
2. Memory allocation high - consider object pooling
3. Performance meets target (<5ms avg)

Skill Used

performance-benchmarking

CI Integration

# Add to CI pipeline
benchmark:
  script:
    - /benchmark src/critical-path/ --type function
    - /benchmark /api/core-endpoints --type api
  allow_failure: false
  rules:
    - if: $CI_PIPELINE_SOURCE == "merge_request"

When to Use Agent Instead

Use the Software Architect agent when:

• Performance requirements need definition
• Architecture changes needed for performance
• Capacity planning required