m10-performance

Systematic approach to identifying and eliminating performance bottlenecks through measurement and targeted optimization. Layer 2: Design Choices What's the bottleneck, and is optimization worth it? Before optimizing: Before optimizing: Have you measured? What's the priority? What's the trade-off? To domain constraints (Layer 3): To implementation (Layer 1): - Emphasizes profiling first (flamegraph, perf, criterion) before optimizing; includes decision table mapping goals (reduce allocations, improve cache, parallelize) to specific implementation patterns - Prioritizes optimization by impact: algorithm choice (10x–1000x), data structure (2x–10x), allocation reduction (2x–5x), cache optimization (1.5x–3x) - Covers common techniques: pre-allocation with with_capacity , avoiding clones via references or Cow , SmallVec for small collections, batch operations - Flags anti-patterns and mistakes: optimizing without profiling, benchmarking in debug mode, LinkedList usage, premature optimization, and hidden clones - Have you measured? (Don't guess) - What's the acceptable performance? - Will optimization add complexity?