m10-performance

name	m10-performance
description	Mastering C++ Performance. Triggers: cache locality, heap allocation, inlining, SIMD, standard layout, string_view, reserve, pmr.

C++ Performance

Core Question

Where is the data?

• Contiguous? (Cache friendly) → std::vector.
• Scattered? (Cache miss) → std::list, std::map.

Thinking Prompt

1. Count the Allocations.

   • Can they be fused? (reserve).
   • Can they be removed? (Stack / SSO).

2. Measure, Don't Guess.

   • Use Google Benchmark or Quick-Bench.

3. Data Layout.

   • struct padding? Reorder members largest to smallest.
   • Pointer chasing? Flatten the graph.

Quick Reference

Technique	Benefit	Implementation
reserve()	Avoid reallocs	Call before loop.
string_view	No copy string	Params.
std::vector	Cache locality	Default container.
PMR	Custom alloc	std::pmr::monotonic_buffer_resource.