AgentSkillsCN

benchmark-functions

衡量函数性能，对比不同实现方案，适用于关键代码路径的优化与调优。

SKILL.md
--- frontmatter
name: benchmark-functions
description: "Measure function performance and compare implementations. Use when optimizing critical code paths."
mcp_fallback: none
category: analysis
tier: 2

Benchmark Functions

Systematically measure function execution time, memory usage, and performance characteristics to identify optimization opportunities.

When to Use

  • •Comparing different algorithm implementations
  • •Measuring performance before/after optimization
  • •Profiling SIMD vs scalar implementations
  • •Establishing performance baselines for CI/CD

Quick Reference

bash
# Python benchmarking with timeit
python3 -m timeit -s 'import module' 'module.function(args)' -n 1000 -r 5

# Mojo benchmarking with built-in timing
mojo run benchmark_script.mojo

Workflow

  1. •Set up benchmarks: Create timing harness with warm-up iterations
  2. •Run measurements: Execute function multiple times, record timing
  3. •Collect statistics: Calculate mean, median, std deviation
  4. •Compare baselines: Compare against previous implementations
  5. •Identify bottlenecks: Pinpoint functions needing optimization

Output Format

Benchmark report:

  • •Function name and parameters tested
  • •Execution time statistics (mean, median, min, max)
  • •Memory usage (if applicable)
  • •Comparison to baseline (improvement percentage)
  • •Iterations and sample size used

References

  • •See profile-code skill for detailed performance profiling
  • •See suggest-optimizations skill for improvement strategies
  • •See CLAUDE.md > Performance for Mojo optimization guidelines