Skip to content

[Apple 5/5] Profile and optimize native performance on every target platform #22

Description

@advatar

Goal

Define and meet measurable frame-time, latency, memory, loading, thermal, and power budgets on Vision Pro, iPhone/iPad, Mac, and Apple TV.

Scope

  • Establish representative device and scene benchmarks.
  • Profile with RealityKit Trace, Metal System Trace, Metal debugger/captures, and Instruments.
  • Track CPU encode time, GPU time, draw calls, triangles, overdraw, texture residency, memory, asset-load latency, and dropped frames.
  • Optimize assets and RealityKit scene updates before adding renderer complexity.
  • Evaluate direct Metal only for a measured bottleneck or capability gap.
  • If direct immersive Metal is justified, prototype it behind an isolated renderer boundary using Compositor Services.

Acceptance criteria

  • Performance budgets and supported device tiers are documented.
  • Automated smoke/performance scenarios cover loading, rotation, selection, clipping, compare, and immersive interaction.
  • Representative hardware meets agreed frame pacing without sustained thermal degradation.
  • Profiling evidence is attached before any decision to replace RealityKit with a custom Metal renderer.
  • Platform-specific degradation paths preserve correctness and interaction quality.

Sequence

This is step 5 of 5 and depends on #19, #20, and #21.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions