Swift Performance Optimization

Language-level performance patterns for memory efficiency, runtime speed, and zero-cost abstractions. Profile first, optimize later — use this skill after Instruments identifies Swift code as the bottleneck.

When to Use

Use this skill when:

• Time Profiler shows Swift code as hotspot
• Excessive memory allocations or retain/release traffic
• Implementing performance-critical algorithms
• Writing framework code with performance requirements
• Optimizing tight loops or frequently called methods

Do NOT use for

• First-step optimization (use performance-profiling first)
• SwiftUI performance (use swiftui-performance)
• Premature optimization

Example Prompts

• "Profiler shows excessive copying — how do I eliminate it?"
• "Retain/release overhead in Time Profiler — how to reduce?"
• "Generic code is slow in hot path"
• "Actor overhead causing UI jank"

What This Skill Provides

Four Principles of Swift Performance

From WWDC 2024-10217 — the organizing mental model for all optimizations:

1. Function Calls – Static vs dynamic dispatch, generic specialization
2. Memory Allocation – Stack vs heap, when each occurs
3. Memory Layout – Contiguous vs pointer-chasing, cache lines
4. Value Copying – COW triggers, defensive copies, ARC traffic

Core Topics

1. Noncopyable Types – Swift 6 ~Copyable for types that should never be copied
2. Copy-on-Write – Use isKnownUniquelyReferenced(), reserveCapacity(), and avoid defensive copies
3. Value vs Reference – Structs under 64 bytes are fast; larger need indirect storage
4. ARC Optimization – unowned is ~2x faster than weak; closure capture costs (escaping vs non-escaping)
5. Generics – Use some over any for static dispatch
6. Collection Performance – ContiguousArray is ~15% faster than Array; InlineArray for fixed sizes
7. Concurrency – Actor hops cost ~100μs; batch calls
8. Memory Layout – Struct padding, exclusivity checks, cache-friendly data structures
9. Span Types – Safe zero-copy memory access with OutputSpan for initialization

Each topic ships with before/after code patterns, measured costs, and a code-review checklist — those live in the skill itself.

Related

• performance-profiling – Use this first to identify bottlenecks
• swift-concurrency – Correctness-focused concurrency
• swiftui-performance – SwiftUI-specific optimization

Resources

WWDC: 2025-312, 2024-10217, 2024-10170, 2021-10216, 2016-416