perf(ui): gate per-pixel clip compares behind a flags bit (#50) #91

Merged

catbot merged 2 commits from claude/issue-50 into master

2026-06-16 18:59:48 +02:00

catbot commented

2026-06-16 18:57:35 +02:00

Member

Summary

uiResolveScreenPixel ran four float compares against hdr.clipRectPx for every pixel of every UI dispatch, even though the clip rect is the full-surface default {0,0,1e9,1e9} in the overwhelmingly common case (issue #50).

This reserves the high bit of the header flags word (UI_FLAG_CLIP) and gates the clip compares on it:

UIRenderer::FillHeader now sets the bit only when the clip rect is narrower than the surface. The decision is extracted into the pure, unit-testable UIRenderer::ClipFlags.
The shader branch is push-constant-uniform across the dispatch, so it never diverges.
When the rect covers the surface, the skipped compares could never have rejected an in-surface pixel — output is pixel-identical.
User flag bits are preserved; only the reserved high bit is touched.

The same gate is mirrored on the WebGPU/WGSL path (additional/dom-webgpu.js) for parity. The frozen-layout contract is respected — this is an additive use of the reserved flags bits.

Testing

New UIClipFlag CPU test drives ClipFlags directly (full-surface vs. each narrowed edge, user-flag preservation, stale-bit clearing) — no Window/GPU needed.
All four standard UI compute shaders + the CustomShader example recompile cleanly with glslang.
crafter-build test — 8 passed.

Resolves #50

🤖 Generated with Claude Code

## Summary `uiResolveScreenPixel` ran four float compares against `hdr.clipRectPx` for every pixel of every UI dispatch, even though the clip rect is the full-surface default `{0,0,1e9,1e9}` in the overwhelmingly common case (issue #50). This reserves the high bit of the header `flags` word (`UI_FLAG_CLIP`) and gates the clip compares on it: - `UIRenderer::FillHeader` now sets the bit **only** when the clip rect is narrower than the surface. The decision is extracted into the pure, unit-testable `UIRenderer::ClipFlags`. - The shader branch is push-constant-uniform across the dispatch, so it never diverges. - When the rect covers the surface, the skipped compares could never have rejected an in-surface pixel — output is **pixel-identical**. - User flag bits are preserved; only the reserved high bit is touched. The same gate is mirrored on the WebGPU/WGSL path (`additional/dom-webgpu.js`) for parity. The frozen-layout contract is respected — this is an additive use of the reserved `flags` bits. ## Testing - New `UIClipFlag` CPU test drives `ClipFlags` directly (full-surface vs. each narrowed edge, user-flag preservation, stale-bit clearing) — no Window/GPU needed. - All four standard UI compute shaders + the CustomShader example recompile cleanly with glslang. - `crafter-build test` — **8 passed**. Resolves #50 🤖 Generated with [Claude Code](https://claude.com/claude-code)

catbot added 1 commit

2026-06-16 18:57:35 +02:00

perf(ui): gate per-pixel clip compares behind a flags bit (#50 ) 65c19ea84a

uiResolveScreenPixel ran four float compares against hdr.clipRectPx for
every pixel of every UI dispatch, even though the clip rect is the
full-surface default {0,0,1e9,1e9} in the overwhelmingly common case.

Reserve the high bit of the header flags word (UI_FLAG_CLIP) and have
FillHeader set it only when the clip rect is narrower than the surface
(extracted into the unit-testable UIRenderer::ClipFlags). The shader
gates the compares on the bit; the branch is push-constant-uniform so it
never diverges. When the rect covers the surface the skipped compares
could never reject an in-surface pixel, so output is pixel-identical.

Mirrored on the WebGPU/WGSL path (dom-webgpu.js) for parity. Adds the
UIClipFlag CPU test covering the gate decision.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

catbot added 1 commit

2026-06-16 18:59:42 +02:00

Merge remote-tracking branch 'origin/master' into claude/issue-50 f2f5a770c0