profile

Profile Code

This skill helps find performance bottlenecks in the code.

Instructions

1. Identify Target:

   • Ask user for the script or function call to profile.

2. Run Profiler:

   • Simple Timer: For quick checks, wrap code in time.perf_counter().
   • cProfile: Run python -m cProfile -s cumulative <script_name.py>.
   • Line Profiler: If detailed line-by-line analysis is needed, suggesting adding @profile decorator and running kernprof -l -v <script_name.py> (requires line_profiler installed).

3. Analyze Output:

   • Look for functions with high cumtime (cumulative time).
   • Look for functions with high call counts (ncalls).

4. Report:

   • Summarize which parts of the code are consuming the most time.
   • Suggest potential optimizations (vectorization, caching, algorithm change).