cc-performance-tuning

Measure-first performance tuning requires running a profiler on the hot path before optimizing code. You collect quantitative metrics, apply a 7-step gated decision tree to guide changes, and produce a violation, warning, and pass table with evidence to document measurable improvements.

Code optimization for high CPU and memory usage should follow a measure-first approach. Profile the hot path to gather quantitative metrics, evaluate the system against a 40-item checklist, and document improvements using a structured violation and warning table.

Yes, this performance tuning process applies to single-process or multi-threaded services, web apps, and batch jobs. It targets slow paths, timeouts, and poor scalability across these contexts by enforcing a measure-first profiling approach before any code optimization.

You should not optimize code without profiling first because performance tuning must be grounded in quantitative metrics. Skipping the profiler and the 7-step gated decision tree risks making changes that fail to resolve the actual bottlenecks or improve scalability.

To document measurable improvements when fixing performance issues, produce a structured violation, warning, and pass table. This report uses evidence collected from profiling to justify code changes and demonstrate the impact on scalability and resource usage.