[perf] Per-glyph-constant SDF AA scale recomputed per pixel (ui-text.comp.glsl)
[perf] Per-glyph-constant SDF AA scale recomputed per pixel (ui-text.comp.glsl)
Merged into #125 — the invRectSize hoist and the SDF-band hoist are the same technique in the same cooperative-load sections of ui-text/ui-fused; one pass through the shaders covers both.
[perf] Per-frame whole-buffer flush of TLAS instance+metadata buffers (RenderingElement3D.cpp)
[perf] Per-frame whole-buffer flush of TLAS instance+metadata buffers (RenderingElement3D.cpp)
Merged into #118 — the TLAS host rebuild and the whole-buffer flush act on the same instance+metadata buffers; one dirty-set drives both the copy and the ranged flush, so they're a single…
[perf] Acquire/present barriers use ALL_COMMANDS scope (Window.cpp)
[perf] Acquire/present barriers use ALL_COMMANDS scope (Window.cpp)
Merged into #115 — the inter-pass and acquire/present ALL_COMMANDS barriers share one fix (derive the real per-pass stage union, including RAY_TRACING_SHADER) and are a single co-dependent…
[perf] No buffer suballocator — vkAllocateMemory per buffer (VulkanBuffer.cppm)
[perf] Win32 StartSync busy-spins one core at 100% when not updating (Window.cpp)
[perf] WebGPU full-buffer readback copies entire allocation even for a small prefix (WebGPUBuffer.cppm)
[perf] UIRenderer::Dispatch heap-allocates a std::vector per custom dispatch (UI-WebGPU.cpp)
[perf] WebGPU GPU-owned instance runs upload one writeBuffer per element (RenderingElement3D-WebGPU.cpp)
[perf] WebGPU metadata flush uploads full 256 KB padded buffer every frame (RenderingElement3D-WebGPU.cpp)
[perf] FontAtlas dirty-rect uses one union bounding box (FontAtlas.cpp)
[perf] DrawInputField recomputes caret X via full-prefix walk every frame (InputField.cpp)
[perf] Unconditional destination imageLoad/imageStore before early-out (ui-quads.comp.glsl)
[perf] Per-glyph-constant SDF AA scale recomputed per pixel (ui-text.comp.glsl)
[perf] Per-pixel divide by item-constant rect size in inner loop (ui-images.comp.glsl)
[perf] Single-lane serial stream compaction stalls 63/64 threads (ui-fused.comp.glsl)
[perf] shapedRuns_ eviction is all-or-nothing clear() at cap (UI-Shared.cpp)
[perf] ShapeText heap-allocates the cache-key std::string on every call, including hits (UI-Shared.cpp)