Document Forward+ measurements and Linux lighting evidence

PLAN.md

@@ -199,16 +199,19 @@ GPU instance record содержит стабильные slot/generation; пл
 
 ### P3. Освещение, тени и temporal reconstruction
 
-**Освещение и тени реализованы до измеренного выбора пути; приёмка этапа ещё
+**Освещение, тени и измеренный выбор пути реализованы; приёмка всего P3 ещё
 открыта.** Есть authored directional/point/spot lights, общий shader ABI для
 Direct и P2, четыре каскада солнца, отдельный 16-face atlas для point/spot,
 видимость каскадеров из shadow views и общий бюджет 4096 caster draws.
 Ранжирование 128 local lights, атомарный отказ от шести point faces и
 unshadowed fallback доступны с диагностикой. На Linux reference GPU Release
-1920×1080 измеренный рост стоимости main raster уже превысил порог для
-Forward+, поэтому depth-free tiled путь 16×16 и повторные измерения входят в
-оставшуюся работу. [Протокол проверки](docs/validation/p3-lighting-2026-09-24/README.md)
-отделяет текущий checkpoint от финальной Linux/Windows приёмки.
+1920×1080 измеренный рост стоимости main raster превысил порог для проверки
+Forward+. Depth-free tiled путь 16×16 прошёл image parity, но полный build +
+raster на плотной контрольной сцене оказался медленнее; `Auto` оставлен на
+forward, явный tiled доступен для локализованных источников и выиграл в
+отдельном сценарии. [Исследование](docs/studies/23-p3-forward-plus-2026-09-24.md)
+и [протокол проверки](docs/validation/p3-lighting-2026-09-24/README.md)
+отделяют этот Linux checkpoint от финальной Windows/temporal приёмки.
 
 Затем: previous transforms, motion vectors, jitter, history rejection и TAA; temporal upscaling — после устойчивого TAA. Проверять тонкую геометрию, движение, disocclusion, camera cut и смену разрешения, сравнивать с режимом без temporal. У cache/pass видны затраты и причины обновления.
 

docs/IMPLEMENTATION.md

@@ -508,6 +508,24 @@ repeats, ten warm-up and thirty measured frames per configuration. It reached
 the agreed Forward+ gate: main-raster overhead at 32 lights was about 0.50 ms
 relative to the matching zero-light case, roughly 30% of that GPU frame;
 64 and 128 lights added about 1.02 and 2.03 ms. Its raw CSV/report are being
-published separately with the exact benchmark revision and driver. A 16×16
-tiled Forward+ path, image parity and before/after build+raster measurement are
-therefore pending. Temporal reconstruction is developed and accepted separately.
+published separately with the exact benchmark revision and driver.
+
+Revision `a0a4e29` adds an explicit depth-free 16×16 tiled Forward+ path. One
+compute invocation tests every submitted point/spot range sphere against a
+tile's four screen-space planes and writes at most 64 stable-order indices.
+An overflowing tile scans the entire submitted list in the fragment shader;
+there is no dropped light. The tile shader has exact reflection validation,
+package/build integration, GPU timing, optional occupancy readback, and
+reload rollback. Direct, GPU frustum and GPU occlusion image tests cover a
+cropped viewport, resize, near-plane light/shadow, and overflow. A 128-light
+localized 1920×1080 forward/tiled capture matched byte for byte.
+
+The [paired Release study](studies/23-p3-forward-plus-2026-09-24.md) measured
+tile build **plus** raster on the same RTX 2080 Ti source revision and shader
+bundle. At 32/64/128 broad overlapping lights it was 0.064/0.122/0.222 ms
+slower; all 8160 tiles overflowed at 128. A separate localized-range scene
+was 0.103/0.214/0.440 ms faster at those counts, with no overflows. There is
+no robust scene/device runtime predictor yet, so `Auto` remains forward and
+`Tiled` is explicit. Linux Debug passed 62 CTests with one window skip;
+pinned Linux SwiftShader passed all six P3 cases. Windows CI on this new
+revision and P3 temporal reconstruction are separate acceptance work.

docs/manual/editor/diagnostics.md

@@ -27,10 +27,16 @@ When GPU timestamps are available, the panel shows sun and local shadow pass
 durations. A zero duration after a disabled sun or sprite-only frame confirms
 that no sun shadow raster ran. The lighting path names the algorithm actually
 used, so compare it with a benchmark's requested mode before interpreting costs.
+For an explicitly tiled frame, the overlay also reports the 16×16 grid size,
+its compute GPU duration, the number of stored light candidates, and how many
+tiles overflowed their 64-index list and scanned all lights. Candidate and
+overflow counts require the diagnostics readback; **unavailable** is distinct
+from a measured zero. The diagnostic copy itself adds work, so close the panel
+before measuring performance.
 See [Lighting](lighting.md) for the 128-light and 16-tile limits.
 
 The Vulkan backend emits `VK_EXT_debug_utils` labels for `SunShadowAtlas`,
-`LocalShadowAtlas`, `ForwardAndUI`, `Readback`, and, when presenting,
+`LocalShadowAtlas`, `LightTileBuild` when tiled, `ForwardAndUI`, `Readback`, and, when presenting,
 `Presentation`. A fallback frame can have no shadow-raster label. A graphics
 capture tool that supports this extension can identify the command-buffer
 regions. Labels remain available without the Khronos validation layer when

docs/manual/editor/lighting.md

@@ -72,6 +72,25 @@ cleared and redrawn each frame; there is no persistent shadow cache yet.
 Sprite-only scenes, a missing sun, and a sun with `casts_shadow: false` skip sun
 shadow raster work.
 
+## Local-light rendering path
+
+The normal `Auto` setting uses the measured forward light scan. It is the
+current default for Editor and Player. A C++ renderer integration can explicitly
+set `RendererConfig::lighting_mode = LightingMode::Tiled` to build depth-free
+16×16 screen-tile lists on a capable Vulkan device. Each tile stores at most
+64 light indices in stable order. If more lights touch a tile, its fragment
+shader scans the complete submitted list, so an overflow never removes
+illumination. The path falls back to forward when no local lights are present
+or the compute/buffer requirements are unavailable. Sprites and UI stay unlit.
+
+This explicit path can help when light ranges occupy small parts of the screen;
+it costs extra work when nearly every light covers nearly every tile. The
+fixed dense benchmark was slower after including tile construction, so there
+is no automatic scene-dependent switch yet. The Player profile reports
+`effective_lighting_path`, tile GPU time and grid size; optional Editor
+diagnostics also report stored candidates and overflowing tiles. See
+[Profiling](profiling.md) and the [measured Forward+ study](https://github.com/emil28092005/Faset_Engine/blob/main/docs/studies/23-p3-forward-plus-2026-09-24.md).
+
 ## Add a point light through MCP
 
 MCP edits the **Editor document**, not entities in a running game. Use

docs/manual/editor/profiling.md

@@ -130,6 +130,15 @@ submitted-light count of zero is a different workload from 128 lights whose
 shadows are disabled. See [Lighting](lighting.md) for the capacity policy and
 [Diagnostics](diagnostics.md) for the Editor counters.
 
+The same sample includes `effective_lighting_path` (`forward` or `tiled`),
+`gpu_light_tiles_ms`, and `light_tile_count`. Stored candidate and overflow
+counts are present only when visibility diagnostics readback was enabled;
+`light_tile_counts_valid: false` means their `null` values are unavailable,
+not zero. The normal `Auto` setting currently resolves to `forward` after the
+fixed dense 1080p benchmark showed that tile construction cost outweighed its
+raster savings. A C++ renderer integration can explicitly request `Tiled` for
+a localized-light scene, then check the actual path before comparing timings.
+
 The fixed-scene benchmark compares 0, 4, 16, 32, 64, and 128 local lights under
 Direct, GPU frustum, and GPU occlusion visibility, with shadows on and off. Its
 wrapper runs three independent 1920×1080 repetitions per configuration, each
@@ -162,6 +171,18 @@ frame** at 32, 64, or 128 lights on the Linux physical reference GPU. The
 states the measured decision and scope. A software Vulkan run checks
 functionality, not physical GPU performance.
 
+For a direct comparison of the two algorithms on the same scene, invoke the
+Release executable twice with `--lighting forward` and `--lighting tiled`,
+using the same `--lights`, `--shadows`, `--visibility`, and output size. The
+default `--light-layout dense` preserves the fixed benchmark scene;
+`--light-layout localized` reduces point-light ranges to 1.75 units as a
+separately labelled workload. Compare `gpu_build_plus_raster_ms`, which includes
+`gpu_light_tiles_ms`, rather than raster time alone. One optional diagnostic
+frame with `--tile-diagnostics on` reports candidate and overflow counts but
+adds a GPU readback, so do not mix it into the timed runs. The
+[Forward+ measurement](https://github.com/emil28092005/Faset_Engine/blob/main/docs/studies/23-p3-forward-plus-2026-09-24.md) retains
+raw frames, shader hashes, and the decision.
+
 ## Current performance scope
 
 The accepted MVP path uses direct draws and CPU culling; P2 adds optional GPU

docs/studies/23-p3-forward-plus-2026-09-24.md

@@ -0,0 +1,105 @@
+# P3 Forward+ experiment: correctness and cost on localized lights
+
+The fixed 1920×1080 P3 benchmark crossed the agreed threshold for trying
+Forward+. A depth-free 16×16 tiled implementation now exists, but the measured
+**build + raster** cost is higher than a full light scan on that benchmark's
+dense lights. `RendererConfig::lighting_mode = Auto` therefore keeps the forward
+path. `Tiled` is an explicit option for scenes whose projected light volumes
+are localized. There is no unmeasured automatic occupancy heuristic.
+
+This is a follow-up to the fixed-scene baseline sweep, which is being merged
+as a separate study. It compares both paths in the same source revision
+`a0a4e29d480ed3344f19bd3565d48668ca913fed`. The baseline's dense
+placement remains the default. An explicit `--light-layout localized` changes
+only point-light range from 8 to 1.75 world units; camera, nine casters,
+receiver, positions, colors, light count, and output size are unchanged. The
+localized fixture is a separate workload, not a replacement for the fixed
+baseline gate.
+
+## Renderer behavior and safety
+
+The compute pass builds up to 64 stable-order light indices per screen tile.
+It tests each world-space range sphere against four clip-space tile planes.
+It does not use depth or reject near-plane intersections. A tile with more than
+64 candidates sets an overflow bit; the fragment shader then scans **all**
+submitted lights for that tile. Zero lights, missing capability, excessive
+buffer size, failed optional allocation, and `Auto` use the forward path. The
+shader contract checks the new compute entry's descriptors and 96-byte push
+constants; Direct and P2 GPU graphics still use materials at set 0, lighting
+at set 1, and GPU scene data at set 2. The tile list is set 1 binding 4 in the
+shared fragment shader. Sprite/UI shading returns before tile reads.
+
+The Linux Vulkan image test compares forward and tiled output in Direct, GPU
+frustum, and GPU occlusion modes, including a cropped scene viewport, near-plane
+crossing point light and shadow, resize, an offscreen light, and 80 coincident
+lights that exceed tile capacity. Every overflowing tile falls back to the full
+list. Shader reload preserves a working tiled pipeline after invalid bytecode
+and recreates it after a valid reload. A separate 1920×1080 capture with 128
+localized lights was byte-identical across both paths; its SHA-256 is in the
+[provenance record](data/p3-forward-plus-provenance-2026-09-24.json).
+
+## Measurement
+
+The device was NVIDIA GeForce RTX 2080 Ti with NVIDIA driver 595.84.0.0,
+Linux Clang Release, Direct visibility, shadows off, 1920×1080. Each mode had
+three independent process runs with ten warm-up and thirty measured frames.
+Forward/tiled run order alternated. The table uses the median of the three
+per-run medians in milliseconds. The tile build column is an actual GPU
+timestamp; `build + raster` also includes post raster if present. The dense
+and localized CSVs contain every one of the 1080 measured frames, with a
+`source_csv` identifier. The executable and all loaded `.spv`/reflection
+SHA-256 values are in the provenance record.
+
+| Light layout | Lights | Forward raster | Tile build | Tiled raster | Tiled build + raster | Tiled change |
+| --- | ---: | ---: | ---: | ---: | ---: | ---: |
+| Dense fixed scene | 32 | 0.5500 | 0.0617 | 0.5527 | 0.6144 | +0.0644 ms (11.7% slower) |
+| Dense fixed scene | 64 | 1.0701 | 0.1177 | 1.0740 | 1.1921 | +0.1220 ms (11.4% slower) |
+| Dense fixed scene | 128 | 2.1172 | 0.2219 | 2.1164 | 2.3388 | +0.2216 ms (10.5% slower) |
+| Localized range 1.75 | 32 | 0.2336 | 0.0555 | 0.0758 | 0.1312 | −0.1025 ms (43.9% faster) |
+| Localized range 1.75 | 64 | 0.4254 | 0.1060 | 0.1057 | 0.2109 | −0.2144 ms (50.4% faster) |
+| Localized range 1.75 | 128 | 0.8094 | 0.2048 | 0.1643 | 0.3691 | −0.4404 ms (54.4% faster) |
+
+At 32 dense lights, the first forward process had a 0.7405 ms run median;
+the other two were 0.5488 and 0.5500 ms. A single paired run would have
+incorrectly suggested a tiled win. The median of three process medians and a
+separate earlier repeat both support the slower dense result. This is why
+`Auto` remains forward despite the localized-scene gain. The total GPU frame
+also includes visibility, shadow fallback, copies, and synchronous readback;
+the table isolates the passes that the optimization changes. For example, at
+32 localized lights the full GPU frame was 1.6494 ms forward and 1.6472 ms
+tiled, essentially unchanged despite lower build + raster cost. At 128 it
+was 2.2788 versus 1.7948 ms.
+
+One diagnostic frame per layout/count copied the tile buffer after the timed
+draw. That copy was **not enabled** in the 1080 performance frames. The grid
+has 8160 tiles and a 64-index capacity per tile.
+
+| Layout | Lights | Stored candidates across tiles | Overflowed tiles |
+| --- | ---: | ---: | ---: |
+| Dense | 32 | 259,896 | 0 |
+| Dense | 64 | 519,792 | 0 |
+| Dense | 128 | 522,240 | 8,160 |
+| Localized | 32 | 38,237 | 0 |
+| Localized | 64 | 76,103 | 0 |
+| Localized | 128 | 152,202 | 0 |
+
+The dense 128 candidate count is capped at 64 × 8160 stored slots; all tiles
+overflow and correctly evaluate all 128 lights in the fragment shader. This
+explains why paying for tile construction cannot help that frame. The localized
+128 scene averages about 19 stored candidates per tile and avoids fallback.
+
+Raw data: [all paired frames](data/p3-forward-plus-ab-2026-09-24.csv),
+[diagnostic frames](data/p3-forward-plus-diagnostics-2026-09-24.csv), and
+[binary/shader provenance](data/p3-forward-plus-provenance-2026-09-24.json).
+
+## Verification and scope
+
+At the implementation revision, Linux Debug built all targets and passed
+62/63 CTests, with the compositor-dependent window lifecycle case skipped and
+no failures. The pinned Linux SwiftShader ICD passed all six P3 cases, including
+the tiled parity/overflow test. The [lighting validation record](../validation/p3-lighting-2026-09-24/README.md)
+retains those logs. These are functional checks on Linux and software Vulkan,
+not physical Windows GPU performance. The A/B numbers apply to one GPU, driver,
+camera, receiver and two synthetic light layouts. They do not establish an
+engine-wide speedup. A measured runtime occupancy predictor and representative
+game scenes are prerequisites before changing `Auto` from forward.

docs/studies/README.md

@@ -1,6 +1,6 @@
 # Исследования для Faset Engine
 
-Обновлено 23.09.2026. Исходники и официальная документация исследовались прежде всего 17.09.2026; затем результаты согласованы с принятой архитектурой. P2 implementation/acceptance добавлены позже и отделены от исходного статического исследования.
+Обновлено 24.09.2026. Исходники и официальная документация исследовались прежде всего 17.09.2026; затем результаты согласованы с принятой архитектурой. P2/P3 implementation/acceptance добавлены позже и отделены от исходного статического исследования.
 
 **Актуальные решения — в [ARCHITECTURE.md](../ARCHITECTURE.md), порядок реализации — в [PLAN.md](../../PLAN.md).** Реализация MVP и проверки идут отдельно: [журнал реализации](../IMPLEMENTATION.md), [результаты проверок](../validation/README.md), [пользовательский Manual](../manual/index.md). Исследования дают обоснования и проверочные сценарии; их статический анализ не является измерением Faset.
 
@@ -41,6 +41,7 @@ Linux/Windows, десктопные 2D/3D, C++ сначала и Lua следу
 - [19 — P2 GPU visibility: протокол приёмки](19-p2-gpu-visibility-acceptance.md): GPU-сценарии, допуски сравнения и методика измерений; отделяет проверку реализации от предложений исследования 15.
 - [20 — P2 GPU visibility: первый benchmark](20-p2-gpu-visibility-benchmark-2026-09-23.md): три запуска, raw CSV, p50/p95 и границы интерпретации на Linux reference GPU.
 - [21 — P2 GPU visibility: оптимизация MainCull](21-p2-gpu-visibility-optimization-2026-09-23.md): отдельные измерения переноса выходных буферов в память GPU и замены CAS-цикла на atomic add.
+- [23 — P3 Forward+: корректность и стоимость](23-p3-forward-plus-2026-09-24.md): сравнение полной цены построения плиток и рисования на плотной и локализованной сценах, raw CSV и точный shader/binary provenance.
 
 ## Происхождение и воспроизводимость
 

docs/studies/data/p3-forward-plus-diagnostics-2026-09-24.csv

@@ -0,0 +1,7 @@
+light_count,light_layout,shadows,visibility,effective_visibility,lighting_path,requested_lighting,build_configuration,run_index,frame,device,driver,commit,width,height,validation_enabled,validation_errors,submitted_local_lights,omitted_local_lights,requested_local_shadow_faces,rendered_local_shadow_faces,dropped_shadow_faces,shadow_atlas_full_drops,shadow_tiles,draw_calls,gpu_bytes,gpu_main_raster_ms,gpu_post_raster_ms,gpu_post_visible,visibility_counters_valid,gpu_sun_shadow_ms,gpu_local_shadow_ms,gpu_shadow_ms,gpu_light_tiles_ms,gpu_build_plus_raster_ms,light_tile_count,light_tile_counts_valid,light_tile_candidate_count,light_tile_overflow_count,gpu_ms,cpu_ms,readback_cpu_ms
+32,dense,off,direct,direct,tiled,tiled,Release,0,0,NVIDIA GeForce RTX 2080 Ti,NVIDIA 595.84.0.0,a0a4e29d480ed3344f19bd3565d48668ca913fed,1920,1080,0,0,32,0,0,0,0,0,0,10,63881248,0.551200,0.000000,0,0,0.000000,0.000000,0.000000,0.061472,0.612672,8160,1,259896,0,3.324384,4.920211,1.106698
+64,dense,off,direct,direct,tiled,tiled,Release,0,0,NVIDIA GeForce RTX 2080 Ti,NVIDIA 595.84.0.0,a0a4e29d480ed3344f19bd3565d48668ca913fed,1920,1080,0,0,64,0,0,0,0,0,0,10,63883808,1.077376,0.000000,0,0,0.000000,0.000000,0.000000,0.117440,1.194816,8160,1,519792,0,3.713568,5.236988,1.014093
+128,dense,off,direct,direct,tiled,tiled,Release,0,0,NVIDIA GeForce RTX 2080 Ti,NVIDIA 595.84.0.0,a0a4e29d480ed3344f19bd3565d48668ca913fed,1920,1080,0,0,128,0,0,0,0,0,0,10,63888928,2.109440,0.000000,0,0,0.000000,0.000000,0.000000,0.221184,2.330624,8160,1,522240,8160,4.435136,6.211166,0.970080
+32,localized,off,direct,direct,tiled,tiled,Release,0,0,NVIDIA GeForce RTX 2080 Ti,NVIDIA 595.84.0.0,a0a4e29d480ed3344f19bd3565d48668ca913fed,1920,1080,0,0,32,0,0,0,0,0,0,10,63881248,0.075328,0.000000,0,0,0.000000,0.000000,0.000000,0.055296,0.130624,8160,1,38237,0,2.355808,3.586245,0.930515
+64,localized,off,direct,direct,tiled,tiled,Release,0,0,NVIDIA GeForce RTX 2080 Ti,NVIDIA 595.84.0.0,a0a4e29d480ed3344f19bd3565d48668ca913fed,1920,1080,0,0,64,0,0,0,0,0,0,10,63883808,0.105248,0.000000,0,0,0.000000,0.000000,0.000000,0.105312,0.210560,8160,1,76103,0,2.445600,3.884127,1.069417
+128,localized,off,direct,direct,tiled,tiled,Release,0,0,NVIDIA GeForce RTX 2080 Ti,NVIDIA 595.84.0.0,a0a4e29d480ed3344f19bd3565d48668ca913fed,1920,1080,0,0,128,0,0,0,0,0,0,10,63888928,0.215040,0.000000,0,0,0.000000,0.000000,0.000000,0.229376,0.444416,8160,1,152202,0,2.585504,4.059668,0.951374

docs/studies/data/p3-forward-plus-provenance-2026-09-24.json

@@ -0,0 +1,33 @@
+{
+  "commit": "a0a4e29d480ed3344f19bd3565d48668ca913fed",
+  "device": "NVIDIA GeForce RTX 2080 Ti",
+  "driver": "NVIDIA 595.84.0.0",
+  "build": "Linux Clang Release",
+  "executable_sha256": "db151a07eeaee04871920f8617f5bcf4c8ac7cb11abf4e36e5595cc526440a36",
+  "shader_bundle_sha256": "60d2abf2f43557f72f39de7a50f4bc92816eb7859c613b2797c0410e72dbaeed",
+  "shader_bundle_hash_method": "SHA256 of UTF-8 canonical JSON mapping sorted shader filenames to per-file SHA256",
+  "shader_files": {
+    "vertexMain.spv": "1ad2631c35d654f48166321ae43d4165043e1b919b14f1dc61caf97b1ada0898",
+    "vertexMain.reflection.json": "b3bdff863d53a0b385e06a36701a83a6ecafe9a93e97281a732cce8ff862c0db",
+    "fragmentMain.spv": "eb65edf62a4ead573f740903dbbd1a61565fe97d234fbe8a09934ce6415bec92",
+    "fragmentMain.reflection.json": "6b62a42ea448d9ae253d78b6b326dfe0d761532ce695b49263689ec649ede36c",
+    "shadowMain.spv": "7643b4d688492b5ee923b9606f0b0e70343ca05fa67673838206119a6f86d8d8",
+    "shadowMain.reflection.json": "d39e2f83c797589296eac9dce139b9ad836a895d6685bc40f79dd2609d687fee",
+    "lightTileMain.spv": "92850edcc5ee5b586a90ce84ee69a98075a6fff3c13db4a57528d9431afe8b68",
+    "lightTileMain.reflection.json": "bf6a9256aaa6d6edacfa1e1c265d7c8ce2c1e4ea4036e0bea078c22fd4b536ca",
+    "gpuVertexMain.spv": "fc9770f304976875f8dd1f6acbe380c77cb06c13276f5596554b3f0b396b31bd",
+    "gpuVertexMain.reflection.json": "09166d80fba7ab4167e97f85125a7c58ab92ea447c1a6a2ab5084e657401dae9",
+    "gpuShadowMain.spv": "2da7e7f96bc7de4e9c8ac365fe0a63f608c7bb26ab3e7bfebfcf88b32e86c71d",
+    "gpuShadowMain.reflection.json": "c9dd844aefb3840f29caad5dfc53986cacc2220f886eda5ad345ebdbb9ff5b1c",
+    "gpuCullMain.spv": "7691360ec55f41d40d43652f506b051cef8b3a3ddcdcc7db61a7d4d97f9e87d8",
+    "gpuCullMain.reflection.json": "996fe5c7e31c6b8f316d8f290b3be0ae7063eb46c500a661e8715e9b822a2598",
+    "gpuHzbMain.spv": "08682362ee2120ed19d260e613285071747c3e436bb09bb6238e46f9cedc2308",
+    "gpuHzbMain.reflection.json": "589818f670881c64f3aba726ee3bc82fe4b9d835a347f99134e0e157dc4a112e",
+    "gpuPostCullMain.spv": "62cb72e356346e1511c6d640a3cecb4a6e5283795338299b357edae5f7f52be4",
+    "gpuPostCullMain.reflection.json": "eed140bbe09c7d14935b688c770261ac0e81677c0974aa811c88ed67ea779b95"
+  },
+  "matrix": "Direct, shadows off, 1920x1080, dense or localized, 32/64/128 lights, 3 repeats, 10 warmup + 30 measured frames; alternating A/B order",
+  "raw_rows": 1080,
+  "localized_128_capture_sha256": "2aa5d408867f7193b7fd83984f2333b188623bda4d083512ede6b8318f11b45c",
+  "localized_128_capture_exact_equal": true
+}

docs/validation/README.md

@@ -5,7 +5,7 @@ These files preserve bounded checks and their inputs. Each record states its sou
 - [MVP acceptance dossier](mvp-acceptance.md): criterion-by-criterion closure, tested revisions and remaining compatibility coverage.
 - [P2 GPU visibility Linux evidence](p2-gpu-visibility-2026-09-23/README.md): Debug/Release GPU acceptance, lavapipe functional checks, relocated Player exports, and explicit platform/performance limits.
 - [P2 pinned SwiftShader compatibility](p2-swiftshader-2026-09-23/README.md): the Windows CI regression, shader capability fix, independent review closure, final native CI and relocated Player evidence.
-- [P3 lighting and shadows](p3-lighting-2026-09-24/README.md): implementation, acceptance matrix, bounded evidence, and remaining Forward+/platform checks; temporal reconstruction is tracked separately.
+- [P3 lighting and shadows](p3-lighting-2026-09-24/README.md): implementation, Forward+ A/B and image parity, acceptance matrix, bounded Linux evidence, and remaining Windows/platform checks; temporal reconstruction is tracked separately.
 - [Windows software Vulkan](windows-software-vulkan-2026-09-18/README.md): fresh native build, 35 tests, launcher/window/MCP workflows and both relocated Release games on SwiftShader.
 - [Checkpoint 5 Linux acceptance](checkpoint5-linux-2026-09-18/README.md): clean offline source build, first Editor launch, exact-candidate standalone games and live Blender checks.
 - [Final Linux source checks](final-linux-2026-09-18/README.md): `4cb8255` integrated test results and both Release games after the asset-relocation correction, including package manifests and standalone captures.

docs/validation/p3-lighting-2026-09-24/README.md

@@ -1,11 +1,11 @@
 # P3 lighting and shadows — acceptance record
 
 This record tracks P3 lighting separately from temporal reconstruction. The
-implementation checkpoint is source revision
-`b191ae0bed77a1544a49504b9a3f07e9a3c691f2` on `feat/p3-lighting`. The
-Manual/record edit itself is documentation-only. A later Forward+ change and
-its measurements require a new revision and validation entry before the lighting
-slice can be called complete. Temporal reconstruction has its own acceptance.
+implementation and measured Forward+ checkpoint is source revision
+`a0a4e29d480ed3344f19bd3565d48668ca913fed` on `feat/p3-lighting`.
+The earlier shadow/benchmark integration checkpoint was `b191ae0`. The
+lighting slice has Linux functional and reference-GPU evidence; Windows CI for
+the new tiled revision is pending. Temporal reconstruction has its own acceptance.
 
 ## Implemented at the checkpoint
 
@@ -31,18 +31,26 @@ slice can be called complete. Temporal reconstruction has its own acceptance.
   requested/effective views, drop reasons, atlas bytes, caster draws, GPU shadow
   durations, and the effective lighting path. Shadow tiles are redrawn every
   frame; no persistent depth cache is claimed.
+- An explicit 16×16 depth-free tiled Forward+ path uses at most 64 light indices
+  per tile and evaluates the complete submitted list on overflow. The compute
+  entry has checked reflection and is included in game builds. `Auto` uses
+  forward: three-run Release measurements found tile build + raster slower on
+  the dense fixed scene. The [paired study](../../studies/23-p3-forward-plus-2026-09-24.md)
+  retains a separate localized-light win and exact binary/shader provenance.
 
 ## Acceptance matrix
 
 | Case | Automated evidence | Current status |
 | --- | --- | --- |
-| Empty, disabled, local-only, multiple sun; schema bounds | `scene_view`, `render_lighting_policy`, `render_offscreen` | Covered by Debug tests at the implementation checkpoint; re-run on final revision |
-| Four cascades, split bounds, subtexel stabilization, offscreen/source-LOD0 caster | `render_lighting_policy`, `render_lighting_sun` | Covered by CPU policy and Linux Vulkan image tests; final-revision runs pending |
-| Spot cone, six point faces and seam, dropped whole point shadow | `render_lighting_local`, `render_lighting_policy` | Covered by Linux Vulkan image and CPU tests; final-revision runs pending |
+| Empty, disabled, local-only, multiple sun; schema bounds | `scene_view`, `render_lighting_policy`, `render_offscreen` | Linux Debug green at `a0a4e29`; integrated revision pending |
+| Four cascades, split bounds, subtexel stabilization, offscreen/source-LOD0 caster | `render_lighting_policy`, `render_lighting_sun` | Linux GPU and pinned SwiftShader P3 green at `a0a4e29` |
+| Spot cone, six point faces and seam, dropped whole point shadow | `render_lighting_local`, `render_lighting_policy` | Linux GPU and pinned SwiftShader P3 green at `a0a4e29` |
 | 128-light/16-face/4096-draw limits, unsupported-atlas fallback | `render_lighting_policy`, `render_offscreen`, `render_lighting_local` | CPU and supported-atlas GPU paths covered; actual unsupported Vulkan device not tested |
-| Direct/GPU frustum/GPU occlusion image parity, P2 reload and 2D/UI independence | `render_lighting_sun`, `render_lighting_local`, `render_shader_reload`, `render_offscreen` | Linux supported-driver paths covered; final-revision runs pending |
-| Driver, profile, real 64×64 benchmark smoke | `render_lighting_benchmark_schema`, `render_lighting_benchmark_smoke`, `player_shutdown_diagnostics` | Focused integration tests passed on `b191ae0`; full raw log pending |
-| 1920×1080 0/4/16/32/64/128 Release sweep, three repeats, both shadow states | `tools/benchmark_p3_lighting.py --sweep` | Baseline measured on Linux physical GPU; raw CSV and post-Forward+ comparison pending publication |
+| Direct/GPU frustum/GPU occlusion image parity, P2 reload and 2D/UI independence | `render_lighting_sun`, `render_lighting_local`, `render_shader_reload`, `render_offscreen` | Linux Debug green at `a0a4e29`; integrated revision pending |
+| Forward+/forward parity, near plane, resize, overflow, and shader reload | `render_lighting_tiled`, `render_shader_reload`, `render_shader_reflection`, `build_schema_publication` | Linux Debug and pinned SwiftShader P3 green at `a0a4e29`; 128-light localized Release captures match exactly |
+| Driver, profile, real 64×64 benchmark smoke | `render_lighting_benchmark_schema`, `render_lighting_benchmark_smoke`, `player_shutdown_diagnostics` | Full Linux Debug green at `a0a4e29` |
+| 1920×1080 0/4/16/32/64/128 Release sweep, three repeats, both shadow states | `tools/benchmark_p3_lighting.py --sweep` | Forward baseline measured; its separate raw study is being integrated |
+| 1920×1080 paired paths, 32/64/128 dense and localized lights | `faset_p3_lighting_benchmark --lighting forward|tiled` | Raw 1080 frames and six diagnostic samples retained in study 23; dense slower, localized faster by build+raster |
 | Windows native build, pinned SwiftShader GPU tests, relocated Release 2D/3D Players | `windows-graphics.yml`, `ci.yml` | New P3 revision has not yet completed Windows CI |
 
 The supported-atlas GPU tests create a renderer with validation requested and
@@ -56,11 +64,12 @@ reference-GPU results cannot establish physical Windows performance.
 
 The P3 CTest registrations are `render_lighting_policy` and
 `render_lighting_benchmark_schema` (CPU), plus `render_lighting_sun`,
-`render_lighting_local`, and `render_lighting_benchmark_smoke` (labelled
-`gpu;p3`). Use `ctest --test-dir build/linux-debug -N -L p3` to confirm those
-five cases exist before running them; an empty test selection is not a pass.
+`render_lighting_local`, `render_lighting_tiled`, and
+`render_lighting_benchmark_smoke` (labelled `gpu;p3`). Use
+`ctest --test-dir build/linux-debug -N -L p3` to confirm those six cases exist
+before running them; an empty test selection is not a pass.
 The Windows full graphics job runs all registered tests, while the native
-Windows CPU job uses `-LE gpu` and therefore excludes the three Vulkan cases.
+Windows CPU job uses `-LE gpu` and therefore excludes the four Vulkan cases.
 
 On the Linux host at `b191ae0`, the [CTest inventory](linux-debug-p3-inventory.txt)
 listed all five cases. The [CPU-only P3 run](linux-debug-cpu-ctest.txt) passed
@@ -71,6 +80,22 @@ physical-GPU baseline was being measured, so it is not a final GPU acceptance
 result. The local host was Linux x86_64, kernel 7.0.0-31-generic; the source
 checkout had documentation changes only during these checks.
 
+At `a0a4e29`, the [full Linux Debug run](linux-debug-tiled-ctest.txt) had
+63 registered cases: 62 passed, no failures, and the compositor-dependent
+window lifecycle case skipped. The [pinned Linux SwiftShader P3 run](linux-swiftshader-tiled-p3-ctest.txt)
+passed all six P3 cases without a skip. The Vulkan image cases requested
+validation and asserted zero reported errors. The RTX 2080 Ti A/B used NVIDIA
+driver 595.84.0.0; study 23 records the executable and shader bundle hashes,
+all raw per-frame timings, tile overflow counts, and exact image equality for
+the localized 128-light capture. Its first dense 32-light forward run was an
+outlier, so the decision uses the median of three process medians rather than
+the apparent win in one paired run.
+
+At the earlier `b191ae0` checkpoint, [GitHub native/manual CI](https://github.com/emil28092005/Faset_Engine/actions/runs/35935899512)
+and [Windows graphics/SwiftShader CI](https://github.com/emil28092005/Faset_Engine/actions/runs/35935899505)
+passed. These jobs did **not** include the new tile shader; Windows CI for
+`a0a4e29` is still required.
+
 ```sh
 cmake --build --preset linux-debug --parallel 2
 ctest --test-dir build/linux-debug -L p3 --no-tests=error --output-on-failure
@@ -83,15 +108,16 @@ The benchmark wrapper retains one raw CSV per run, a merged CSV, and a summary.
 It rejects visibility fallback, missing GPU timestamps, missing lights, duplicate
 frames, and validation errors. An offscreen capture's `cpu_ms` includes GPU wait
 and readback; it is not thread CPU time. The exact Release benchmark revision,
-driver, CSV paths, before/after Forward+ gate, Linux SwiftShader results, final
-Debug/Release CTest logs, and Windows Actions links will be added after those
-checks run. Do not use this provisional record as a P3 completion claim.
+driver and paired path data are retained in study 23. Release full CTest,
+final integrated Windows Actions and relocated Player checks still need to be
+added before this is a complete P3 lighting acceptance record.
 
 ## Limits carried forward
 
-The current checkpoint scans all submitted lights in each mesh fragment; the
-measured Forward+ threshold was reached on the Linux reference GPU, so a bounded
-tiled path is in progress. Transparent/game UI and sprites keep their existing
+The default path scans all submitted lights in each mesh fragment; the
+measured Forward+ threshold prompted a bounded tiled implementation. `Auto`
+still uses forward because this dense fixed workload was slower after tile
+construction. Transparent/game UI and sprites keep their existing
 ordering and unlit behavior. The atlas caps are fixed budgets, not adaptive
 quality settings, and shadow depth is redrawn each frame. The renderer still
 performs synchronous framebuffer readback. No broad scene/driver matrix or

docs/validation/p3-lighting-2026-09-24/linux-debug-tiled-ctest.txt

@@ -0,0 +1,140 @@
+Test project /home/emil/Desktop/.worktrees/Faset_Engine-p3-lighting/build/linux-debug
+      Start  1: authoring
+ 1/63 Test  #1: authoring ..........................   Passed    0.06 sec
+      Start  2: runtime_contracts
+ 2/63 Test  #2: runtime_contracts ..................   Passed    0.12 sec
+      Start  3: assets_pipeline
+ 3/63 Test  #3: assets_pipeline ....................   Passed    0.14 sec
+      Start  4: assets_blender_bundle
+ 4/63 Test  #4: assets_blender_bundle ..............   Passed    0.08 sec
+      Start  5: lua_contracts
+ 5/63 Test  #5: lua_contracts ......................   Passed    0.11 sec
+      Start  6: lua_safety_contracts
+ 6/63 Test  #6: lua_safety_contracts ...............   Passed    0.21 sec
+      Start  7: render_lighting_sun
+ 7/63 Test  #7: render_lighting_sun ................   Passed    0.92 sec
+      Start  8: render_lighting_local
+ 8/63 Test  #8: render_lighting_local ..............   Passed    0.87 sec
+      Start  9: render_lighting_tiled
+ 9/63 Test  #9: render_lighting_tiled ..............   Passed    2.75 sec
+      Start 10: render_lighting_policy
+10/63 Test #10: render_lighting_policy .............   Passed    0.04 sec
+      Start 11: render_graph
+11/63 Test #11: render_graph .......................   Passed    0.00 sec
+      Start 12: render_offscreen
+12/63 Test #12: render_offscreen ...................   Passed    0.43 sec
+      Start 13: render_sprite_alpha
+13/63 Test #13: render_sprite_alpha ................   Passed    0.36 sec
+      Start 14: render_shader_reload
+14/63 Test #14: render_shader_reload ...............   Passed    4.49 sec
+      Start 15: render_gpu_shader_contract
+15/63 Test #15: render_gpu_shader_contract .........   Passed    0.06 sec
+      Start 16: render_shader_reflection
+16/63 Test #16: render_shader_reflection ...........   Passed    3.63 sec
+      Start 17: render_window_lifecycle
+17/63 Test #17: render_window_lifecycle ............***Skipped   4.02 sec
+      Start 18: render_lighting_benchmark_schema
+18/63 Test #18: render_lighting_benchmark_schema ...   Passed    4.02 sec
+      Start 19: render_lighting_benchmark_smoke
+19/63 Test #19: render_lighting_benchmark_smoke ....   Passed    0.44 sec
+      Start 20: visibility_policy
+20/63 Test #20: visibility_policy ..................   Passed    0.01 sec
+      Start 21: gpu_visibility
+21/63 Test #21: gpu_visibility .....................   Passed    0.90 sec
+      Start 22: render_gpu_empty
+22/63 Test #22: render_gpu_empty ...................   Passed    0.59 sec
+      Start 23: render_gpu_capacity
+23/63 Test #23: render_gpu_capacity ................   Passed    0.65 sec
+      Start 24: render_gpu_dense
+24/63 Test #24: render_gpu_dense ...................   Passed    0.64 sec
+      Start 25: render_gpu_door
+25/63 Test #25: render_gpu_door ....................   Passed    0.60 sec
+      Start 26: render_gpu_shadow
+26/63 Test #26: render_gpu_shadow ..................   Passed    0.60 sec
+      Start 27: render_gpu_cut
+27/63 Test #27: render_gpu_cut .....................   Passed    0.60 sec
+      Start 28: render_gpu_resize
+28/63 Test #28: render_gpu_resize ..................   Passed    0.63 sec
+      Start 29: render_gpu_lod
+29/63 Test #29: render_gpu_lod .....................   Passed    0.66 sec
+      Start 30: render_gpu_teleport
+30/63 Test #30: render_gpu_teleport ................   Passed    0.61 sec
+      Start 31: render_gpu_near
+31/63 Test #31: render_gpu_near ....................   Passed    0.61 sec
+      Start 32: render_gpu_lifecycle
+32/63 Test #32: render_gpu_lifecycle ...............   Passed    1.32 sec
+      Start 33: render_gpu_views
+33/63 Test #33: render_gpu_views ...................   Passed    0.59 sec
+      Start 34: render_gpu_projection
+34/63 Test #34: render_gpu_projection ..............   Passed    0.57 sec
+      Start 35: render_gpu_open_sequence
+35/63 Test #35: render_gpu_open_sequence ...........   Passed    1.80 sec
+      Start 36: render_gpu_transparent
+36/63 Test #36: render_gpu_transparent .............   Passed    0.62 sec
+      Start 37: player_scene_contracts
+37/63 Test #37: player_scene_contracts .............   Passed    0.03 sec
+      Start 38: player_shutdown_diagnostics
+38/63 Test #38: player_shutdown_diagnostics ........   Passed    1.53 sec
+      Start 39: lua_cli_contracts
+39/63 Test #39: lua_cli_contracts ..................   Passed    0.13 sec
+      Start 40: lua_player_reload
+40/63 Test #40: lua_player_reload ..................   Passed    2.45 sec
+      Start 41: editor_mcp
+41/63 Test #41: editor_mcp .........................   Passed    0.01 sec
+      Start 42: process_and_cook
+42/63 Test #42: process_and_cook ...................   Passed    0.32 sec
+      Start 43: build_schema_publication
+43/63 Test #43: build_schema_publication ...........   Passed    9.27 sec
+      Start 44: editor_plugins
+44/63 Test #44: editor_plugins .....................   Passed    0.03 sec
+      Start 45: editor_session_settings
+45/63 Test #45: editor_session_settings ............   Passed    0.02 sec
+      Start 46: ui_widgets
+46/63 Test #46: ui_widgets .........................   Passed    0.12 sec
+      Start 47: ui_render
+47/63 Test #47: ui_render ..........................   Passed    0.48 sec
+      Start 48: editor_ui_import_conflicts
+48/63 Test #48: editor_ui_import_conflicts .........   Passed    0.82 sec
+      Start 49: editor_ui_project_settings
+49/63 Test #49: editor_ui_project_settings .........   Passed    0.97 sec
+      Start 50: editor_ui_reload
+50/63 Test #50: editor_ui_reload ...................   Passed    4.04 sec
+      Start 51: editor_ui_launcher
+51/63 Test #51: editor_ui_launcher .................   Passed    0.84 sec
+      Start 52: editor_ui_templates
+52/63 Test #52: editor_ui_templates ................   Passed    2.48 sec
+      Start 53: editor_ui_gizmos
+53/63 Test #53: editor_ui_gizmos ...................   Passed    0.92 sec
+      Start 54: editor_ui_authoring
+54/63 Test #54: editor_ui_authoring ................   Passed    1.71 sec
+      Start 55: editor_mcp_stdio
+55/63 Test #55: editor_mcp_stdio ...................   Passed    0.27 sec
+      Start 56: editor_gui_mcp
+56/63 Test #56: editor_gui_mcp .....................   Passed    4.65 sec
+      Start 57: core
+57/63 Test #57: core ...............................   Passed    0.16 sec
+      Start 58: tutorial_moving
+58/63 Test #58: tutorial_moving ....................   Passed    0.02 sec
+      Start 59: tutorial_following
+59/63 Test #59: tutorial_following .................   Passed    0.01 sec
+      Start 60: tutorial_spawning
+60/63 Test #60: tutorial_spawning ..................   Passed    0.02 sec
+      Start 61: tutorial_physics
+61/63 Test #61: tutorial_physics ...................   Passed    0.06 sec
+      Start 62: playable_2d
+62/63 Test #62: playable_2d ........................   Passed    0.40 sec
+      Start 63: playable_3d
+63/63 Test #63: playable_3d ........................   Passed    0.93 sec
+
+100% tests passed, 0 tests failed out of 63
+
+Label Time Summary:
+gpu       =  47.20 sec*proc (35 tests)
+p2        =  12.02 sec*proc (16 tests)
+p3        =   9.05 sec*proc (6 tests)
+window    =   8.67 sec*proc (2 tests)
+
+Total Test time (real) =  67.51 sec
+
+The following tests did not run:
+	 17 - render_window_lifecycle (Skipped)

docs/validation/p3-lighting-2026-09-24/linux-swiftshader-tiled-p3-ctest.txt

@@ -0,0 +1,21 @@
+Test project /home/emil/Desktop/.worktrees/Faset_Engine-p3-lighting/build/linux-debug
+    Start  7: render_lighting_sun
+1/6 Test  #7: render_lighting_sun ................   Passed    5.14 sec
+    Start  8: render_lighting_local
+2/6 Test  #8: render_lighting_local ..............   Passed    4.76 sec
+    Start  9: render_lighting_tiled
+3/6 Test  #9: render_lighting_tiled ..............   Passed   14.30 sec
+    Start 10: render_lighting_policy
+4/6 Test #10: render_lighting_policy .............   Passed    0.03 sec
+    Start 18: render_lighting_benchmark_schema
+5/6 Test #18: render_lighting_benchmark_schema ...   Passed    4.04 sec
+    Start 19: render_lighting_benchmark_smoke
+6/6 Test #19: render_lighting_benchmark_smoke ....   Passed    1.23 sec
+
+100% tests passed, 0 tests failed out of 6
+
+Label Time Summary:
+gpu    =  25.42 sec*proc (4 tests)
+p3     =  29.50 sec*proc (6 tests)
+
+Total Test time (real) =  29.50 sec

docs/validation/p3-lighting-2026-09-24/strict-mkdocs-tiled.txt

@@ -0,0 +1,20 @@
+warning: An executable named `mkdocs` is not provided by package `mkdocs-material` but is available via the dependency `mkdocs`. Consider using `uvx --from mkdocs mkdocs` instead.
+
+ │  ⚠  Warning from the Material for MkDocs team
+ │
+ │  MkDocs 2.0, the underlying framework of Material for MkDocs,
+ │  will introduce backward-incompatible changes, including:
+ │
+ │  × All plugins will stop working – the plugin system has been removed
+ │  × All theme overrides will break – the theming system has been rewritten
+ │  × No migration path exists – existing projects cannot be upgraded
+ │  × Closed contribution model – community members can't report bugs
+ │  × Currently unlicensed – unsuitable for production use
+ │
+ │  Our full analysis:
+ │
+ │  https://squidfunk.github.io/mkdocs-material/blog/2026/02/18/mkdocs-2.0/
+
+INFO    -  Cleaning site directory
+INFO    -  Building documentation to directory: /home/emil/Desktop/.worktrees/Faset_Engine-p3-lighting/build/manual
+INFO    -  Documentation built in 0.76 seconds