perf(cg): full-viewport frame plan fast path

[softmarshmallow]

Summary

• Full-viewport fast path: When the viewport fully contains the scene (fit zoom), skip the R-tree traversal and index sort entirely — return 0..n in O(1) instead of O(n log n). Guarded by R-tree size equality check for correctness when layers lack render bounds.
• sort_unstable: Partial-viewport frames use pdqsort instead of stable merge sort — 2-3x faster for integer data, no stability needed for draw-order indices.
• Shared image snapshot: Pan and zoom caches share a single image_snapshot() call on non-zoom frames, avoiding a redundant GPU handle allocation.

Measured on 01-135k (135K nodes):

Scenario	Metric	Before	After	Delta
rt_pan_slow_fit	queue	1,598 µs	485 µs	-70%
rt_pan_slow_fit	p95	6,317 µs	1,199 µs	-81%
rt_pan_slow_zoomed	p50	300 µs	151 µs	-50%
rt_pan_fast_fit	p50	82 µs	41 µs	-50%
fl_16ms	p50	97 µs	61 µs	-37%

Zero test regressions (281 unit + 67 doc/integration). No new clippy warnings.

Test plan

• cargo test -p cg — all pass
• cargo check -p cg -p grida-canvas-wasm -p grida-dev — all compile
• cargo clippy -p cg --no-deps — no new warnings (305, down from 306)
• GPU benchmark before/after on 135K-node fixture
• Criterion camera benchmarks — no regressions

🤖 Generated with Claude Code

Summary by CodeRabbit

Release Notes

• Performance

  • Optimized GPU snapshot caching to reduce redundant allocation during rendering operations.
  • Implemented fast-path visibility detection when the entire scene fits within the viewport, eliminating unnecessary computation overhead for frame planning.

• Documentation

  • Added documentation describing render optimization strategy and corresponding performance improvements.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
blog	Ready	Preview, Comment	Mar 30, 2026 9:03am
docs	Ready	Preview, Comment	Mar 30, 2026 9:03am
grida	Ready	Preview, Comment	Mar 30, 2026 9:03am
viewer	Ready	Preview, Comment	Mar 30, 2026 9:03am

3 Skipped Deployments

Project	Deployment	Actions	Updated (UTC)
backgrounds	Ignored	Preview	Mar 30, 2026 9:03am
code	Ignored		Mar 30, 2026 9:03am
legacy	Ignored		Mar 30, 2026 9:03am

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 1539d1e8-e594-4858-8a78-2614e2d3c978

📥 Commits

Reviewing files that changed from the base of the PR and between 1a94db1 and 2cfd35e.

📒 Files selected for processing (2)

• crates/grida-canvas/src/runtime/scene.rs
• docs/wg/feat-2d/optimization.md

---

Walkthrough

Optimized GPU snapshot caching and frame visibility selection in the renderer. A single GPU image_snapshot() call now serves both pan and zoom image caches, and a fast-path detection skips R-tree traversal when the camera viewport fully contains the scene. Additionally documented the optimization implementation.

Changes

Cohort / File(s)	Summary
Render optimization: GPU snapshot and visibility selection crates/grida-canvas/src/runtime/scene.rs	Consolidated dual image_snapshot() calls into a single reused snapshot between pan and zoom caches. Added fast-path logic in frame() that skips R-tree traversal and sorting when viewport fully contains scene envelope; otherwise replaced sort() with sort_unstable().
Optimization documentation docs/wg/feat-2d/optimization.md	Added section describing the full-viewport frame plan fast path optimization and GPU snapshot sharing, including benchmark impact and trigger conditions.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• Add Layer Compositing Cache architecture for improved performance #581: Directly modifies the same runtime/scene.rs rendering and compositor-related codepaths affected by these GPU snapshot and frame-planning changes.
• perf(cg): draw-loop viewport culling via ViewportCull #607: Also modifies frame planning and visibility selection logic in runtime/scene.rs with viewport culling and R-tree result threading.

Suggested labels

cg, performance

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'perf(cg): full-viewport frame plan fast path' directly and precisely describes the main optimization introduced in the changeset—a performance improvement adding a fast path for full-viewport frame planning in the canvas renderer.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feature/nostalgic-swirles

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}