Foundation Set 9: Performance & Profiling (25 Q&A)

⚡ Performance & Profiling - 25 Quick Q&A

Q1: First step when app feels slow?

A: Measure before optimizing; reproduce with profiler.

Q2: Which tool for CPU hotspots?

A: Time Profiler.

Q3: Which tool for memory leaks?

A: Leaks + Memory Graph.

Q4: Janky scrolling common cause?

A: Heavy work on main thread during cell rendering.

Q5: Rule for main thread?

A: UI only; move expensive compute/IO off main.

Q6: Why image decoding causes stutter?

A: Decode/compression cost at render time.

Q7: Optimization for large image lists?

A: Predecode, resize, cache thumbnails.

Q8: What is FPS target?

A: 60fps baseline (or 120fps on ProMotion devices where needed).

Q9: What is overdraw?

A: Same pixel rendered multiple times due to layered views.

Q10: How to reduce overdraw?

A: Flatten hierarchy, avoid unnecessary transparency.

Q11: Symptoms of retain cycle?

A: Objects never deinit, memory grows over time.

Q12: Typical retain cycle pattern?

A: Closures strongly capturing self.

Q13: Fix?

A: [weak self] or ownership redesign.

Q14: What’s allocation churn?

A: Frequent object create/destroy causing CPU/memory pressure.

Q15: How to reduce churn?

A: Reuse objects, pools, avoid redundant formatting/parsing.

Q16: Why pagination helps list performance?

A: Bounds memory and network load.

Q17: Startup optimization first lever?

A: Defer non-critical initialization.

Q18: Cold start vs warm start?

A: Cold: fresh process launch; warm: app already resident.

Q19: Why signposts useful?

A: Correlate code phases with profiler timelines.

Q20: p50 vs p95 for perf metrics?

A: p95 reveals tail pain users actually feel.

Q21: How to profile network impact?

A: Measure request count, payload size, serialization cost.

Q22: How to avoid layout thrashing?

A: Cache heights, reduce repeated constraint recalculation.

Q23: When not to micro-optimize?

A: If bottleneck is unknown or impact is negligible.

Q24: Interview framing for optimization work?

A: Baseline -> hypothesis -> experiment -> measurable result.

Q25: One-line principle?

A: Performance engineering is systems thinking + disciplined measurement.