Fr0zka/VoxelForge
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
5aa285c3ef32ec8c6034eb288b3fb456837e6113
VoxelForge/Standalone/Research/OwnerShapeWorld/gpu_field_port
T
History
Fr0zkaandClaude Opus 5.5 5aa285c3ef wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers 
Luna: near skirts were CPU-only by implementation gap, not correctness. Ported with
near-resolution refinement and overflow fallback; near-probe preparation split out so
the skirt kernels no longer inline the near field (first attempt: two fxc /O3 compiles
ran 10.5 h and were stopped). Shape compile proven identical at W1/W8/W16.
-SkipCpuDetailBand retired (154/640 skirt audit failures, +174 ms planning).
Audits: hashes W1/W8/W16, walk 5,642/5,642, whole-near 960 tiles + 640 skirt faces
0 sign/topology diffs (one 0.001 cm vertex-key delta), far/thin 0 diffs, both
survivals pass with 2,074 candidates at 7.00 / 7.02 GiB (was ~6.2), ten captures pass.

My runs outside the sandbox (Step28.exe). NOTE the machine was ~25% slower than last
night for every build (step 27 CPU full machine 0.97-1.02 s at night, 1.22-1.29 s now):
  0xFF00: CPU 1.337 / 1.291 / 1.295 s; GPU miss 1.337 s;
          GPU cached 1.663 / 1.424 / 1.617 s (GPU tiles 1,427 / 2,072 / 1,222)
  0xFFFF: CPU 1.221 / 1.231 / 1.236 s (step 27 same session 1.285 / 1.222 s)
          GPU cached 1.582 / 1.550 / 1.416, 1.635 / 1.441 / 1.470 / 1.445 s
          (GPU tiles vary 593-2,074; timestamps up to ~470 ms)
=> GPU route now slower than the CPU route measured alongside it. Step 27 GPU control
   was not clean (both exes load the same precompiled .cso files).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MmSoF64BbRFmk9C3p1VsBJ
2026-10-08 17:07:40 +02:00
..
precompiled
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
AnalyzeStep6NearAudit.ps1
wip(standalone): add missing GPU skirt shader binaries and step 6-8 scripts/report
2026-10-07 06:14:04 +02:00
AnalyzeStep15Timeline.py
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
BuildGpuFieldPort.ps1
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
BuildGpuFieldStep6NearRunner.ps1
wip(standalone): add missing GPU skirt shader binaries and step 6-8 scripts/report
2026-10-07 06:14:04 +02:00
BuildGpuFieldViewer.ps1
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
BuildGpuFirstSubmitRepro.ps1
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
CompareFarCpuHashes.ps1
wip(standalone): GPU field step 2 -- far-field shapes/strata/relief in shared C++/HLSL, GPU field+mesh
2026-10-06 08:02:02 +02:00
CompareStep6Captures.py
wip(standalone): add missing GPU skirt shader binaries and step 6-8 scripts/report
2026-10-07 06:14:04 +02:00
CompareStep7Captures.py
wip(standalone): add missing GPU skirt shader binaries and step 6-8 scripts/report
2026-10-07 06:14:04 +02:00
CompareStep8Captures.py
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
CompareStep8LodIsolations.py
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
CompareStep9Captures.py
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
CompareStep10Captures.py
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
CompareStep11Captures.py
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
CompareStep12Captures.py
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
CompareStep13Captures.py
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
CompareStep14Captures.py
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
CompareStep15Captures.py
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
CompareStep17TileControls.py
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
CompileStep6GpuShaders.ps1
wip(standalone): add missing GPU skirt shader binaries and step 6-8 scripts/report
2026-10-07 06:14:04 +02:00
CompileStep14Shader.ps1
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
GpuFarViewer.h
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
GpuFieldPort.cpp
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
GpuFirstSubmitRepro.cpp
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
LUNA_DIALOGUE.md
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
REPORT.md
wip(standalone): GPU step 3 -- GPU decides far rock/air, sparse CPU exceptions, hybrid viewer (H)
2026-10-06 09:14:22 +02:00
RunGpuFarAgreement.ps1
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
RunGpuFarViewer.ps1
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
RunGpuFieldPort.ps1
wip(standalone): GPU field step 2 -- far-field shapes/strata/relief in shared C++/HLSL, GPU field+mesh
2026-10-06 08:02:02 +02:00
RUNNERS.md
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
RunStep6CaptureMatrix.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6ColdInterleaved.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6FarAgreement.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6FarPacked.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6GpuCrossing.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6NearAgreement.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6NearDiag.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6NearDiagFull.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6NearMismatchProbe.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep6NearPacked.ps1
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
RunStep7CaptureMatrix.ps1
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
RunStep7ColdInterleaved.ps1
wip(standalone): GPU step 7 partial -- parity check deferred off cold path (stopped early)
2026-10-06 20:17:37 +02:00
RunStep7Process.ps1
wip(standalone): GPU step 7 partial -- parity check deferred off cold path (stopped early)
2026-10-06 20:17:37 +02:00
RunStep8CaptureMatrix.ps1
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
RunStep8ColdInterleaved.ps1
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
RunStep8PresentControls.ps1
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
RunStep8PresentVisibilityControls.ps1
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
RunStep9NormalWindowInterleaved.ps1
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
RunStep10CaptureMatrix.ps1
wip(standalone): GPU step 10 -- first-submit stall bisected to cold shader use under 8 workers
2026-10-07 05:38:14 +02:00
RunStep10CpuReference.ps1
wip(standalone): GPU step 10 -- first-submit stall bisected to cold shader use under 8 workers
2026-10-07 05:38:14 +02:00
RunStep10FinalInterleaved.ps1
wip(standalone): GPU step 10 -- first-submit stall bisected to cold shader use under 8 workers
2026-10-07 05:38:14 +02:00
RunStep10GpuSurvival.ps1
wip(standalone): GPU step 10 -- first-submit stall bisected to cold shader use under 8 workers
2026-10-07 05:38:14 +02:00
RunStep10WarmupBisect.ps1
wip(standalone): GPU step 10 -- first-submit stall bisected to cold shader use under 8 workers
2026-10-07 05:38:14 +02:00
RunStep11AllCpuAffinity.ps1
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
RunStep11CaptureMatrix.ps1
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
RunStep11CpuReference.ps1
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
RunStep11FinalInterleaved.ps1
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
RunStep11GpuSurvival.ps1
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
RunStep11ShaderFirstUse.ps1
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
RunStep11WarmupControls.ps1
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
RunStep12CaptureMatrix.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12ConnectorLoopMeasurement.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12CpuReference.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12FinalInterleaved.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12FinalShaderFirstUse.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12GpuCrossing.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12GpuSurvival.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12PostPoolInterleaved.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12ShaderBaseline.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12ShaderFirstUse.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12SkirtLoopMeasurement.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep12WholeNearAgreement.ps1
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
RunStep13CaptureMatrix.ps1
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
RunStep13CpuReference.ps1
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
RunStep13FinalInterleaved.ps1
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
RunStep13GpuSurvival.ps1
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
RunStep13ShaderFirstUse.ps1
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
RunStep14CaptureMatrix.ps1
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
RunStep14CpuReference.ps1
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
RunStep14FinalInterleaved.ps1
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
RunStep14GpuSurvival.ps1
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
RunStep14ShaderFirstUse.ps1
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
RunStep14WarmupProfile.ps1
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
RunStep15CaptureMatrix.ps1
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
RunStep15CpuReference.ps1
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
RunStep15GpuSurvival.ps1
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
RunStep15Timeline.ps1
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
RunStep16CpuDrainControl.ps1
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
RunStep16CpuPresentControl.ps1
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
RunStep16FairRace.ps1
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
RunStep16GpuWarmupControl.ps1
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
RunStep17TileControls.ps1
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
StartGpuFarViewer.ps1
wip(standalone): GPU step 4 -- batched far generation, connector panels on GPU; far survival passes
2026-10-06 11:01:03 +02:00
step3_cpu_far.png
wip(standalone): GPU step 3 -- GPU decides far rock/air, sparse CPU exceptions, hybrid viewer (H)
2026-10-06 09:14:22 +02:00
step3_gpu_far.png
wip(standalone): GPU step 3 -- GPU decides far rock/air, sparse CPU exceptions, hybrid viewer (H)
2026-10-06 09:14:22 +02:00
step6_capture_comparison.md
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_chamber.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_far.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_passage_floor_closeup.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_passage.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_remote_yaw0.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_remote_yaw90.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_remote_yaw180.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_remote_yaw270.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_tunnel_bend.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_cpu_tunnel.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_chamber.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_far.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_passage_floor_closeup.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_passage_normalfix.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_passage.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_remote_yaw0.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_remote_yaw90.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_remote_yaw180.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_remote_yaw270.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_tunnel_bend.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step6_gpu_tunnel.png
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
step7_capture_comparison.md
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_chamber.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_far.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_passage_floor_closeup.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_passage.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_remote_yaw0.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_remote_yaw90.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_remote_yaw180.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_remote_yaw270.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_tunnel_bend.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_cpu_tunnel.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_chamber.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_far.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_passage_floor_closeup.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_passage.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_remote_yaw0.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_remote_yaw90.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_remote_yaw180.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_remote_yaw270.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_tunnel_bend.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_gpu_tunnel.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_probe_cpu_remote_yaw0.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_probe_gpu_remote_yaw0.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_seam_cpu_remote_yaw0.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step7_seam_gpu_remote_yaw0.png
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
step8_capture_comparison.md
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_clamped_capture_comparison.md
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_chamber.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_far.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_passage_floor_closeup.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_passage.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_remote_yaw0.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_remote_yaw90.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_remote_yaw180.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_remote_yaw270.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_tunnel_bend.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_cpu_tunnel.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_chamber.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_far.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_passage_floor_closeup.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_passage.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_protected_remote_yaw0.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_protected_remote_yaw90.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_remote_yaw0.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_remote_yaw90_direct_audit.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_remote_yaw90_drawargs_full.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_remote_yaw90_drawargs.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_remote_yaw90.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_remote_yaw180.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_remote_yaw270.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_tunnel_bend.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_gpu_tunnel.png
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_lod_isolation_comparison.md
wip(standalone): GPU step 8 partial -- GPU skirts, first-present hold; remote gap open
2026-10-07 02:19:38 +02:00
step8_report.md
wip(standalone): add missing GPU skirt shader binaries and step 6-8 scripts/report
2026-10-07 06:14:04 +02:00
step9final_capture_comparison.md
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_cpu_remote_yaw0.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_cpu_remote_yaw90.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_cpu_remote_yaw180.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_cpu_remote_yaw270.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_gpu_remote_yaw0.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_gpu_remote_yaw90.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_gpu_remote_yaw180.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_gpu_remote_yaw270.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_lod3_remote_cpu.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_lod3_remote_gpu.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_lod4_remote_cpu.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_lod4_remote_gpu.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_remote_cpu.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step9final_remote_gpu.png
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
step10_capture_comparison.md
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
step11_capture_comparison.md
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
step12_capture_comparison.md
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
step13_capture_comparison.md
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
step14_capture_comparison.md
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
step15_capture_comparison.md
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
step15_timeline_fast_cpu_tiles.csv
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
step15_timeline_fast_frames.csv
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
step15_timeline_fast_gpu_batches.csv
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
step15_timeline_paced_cpu_tiles.csv
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
step15_timeline_paced_frames.csv
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
step15_timeline_paced_gpu_batches.csv
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
step16_capture_comparison.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
step17_final_capture_comparison.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
step17_final_r29_capture_comparison.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
step17_final_r29_tilecontrol_comparison.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
step17_r26_tilecontrol_comparison.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
step17_tilecontrol_comparison.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
step18_final_capture_comparison.md
wip(standalone): GPU step 18 -- warm-up marker: hybrid now wins cached and ties first launch
2026-10-07 21:07:05 +02:00
step18_tilecontrol_comparison.md
wip(standalone): GPU step 18 -- warm-up marker: hybrid now wins cached and ties first launch
2026-10-07 21:07:05 +02:00
step19_capture_capture_comparison.md
wip(standalone): GPU step 19 -- streamed GPU input export; RUNNERS.md
2026-10-07 22:09:29 +02:00
step19_final_capture_comparison.md
wip(standalone): GPU step 19 -- streamed GPU input export; RUNNERS.md
2026-10-07 22:09:29 +02:00
step19_warm_capture_comparison.md
wip(standalone): GPU step 19 -- streamed GPU input export; RUNNERS.md
2026-10-07 22:09:29 +02:00
step20final_capture_comparison.md
wip(standalone): GPU step 20 -- detail band batched by 8 columns; gap instrumented
2026-10-07 22:35:22 +02:00
step21_capture_comparison.md
wip(standalone): GPU step 21 -- GPU input export shares the 8 workers (3:1 priority); no wall gain
2026-10-07 23:20:42 +02:00
step22_capture_comparison.md
wip(standalone): GPU step 22 -- startup multi-batch pump, pool before handoff: cached 1.12-1.20 s
2026-10-08 00:00:06 +02:00
step22_final_capture_comparison.md
wip(standalone): GPU step 22 -- startup multi-batch pump, pool before handoff: cached 1.12-1.20 s
2026-10-08 00:00:06 +02:00
step23_final_capture_comparison.md
wip(standalone): GPU step 23 -- skip empty connector-ring dispatch; cached 1.07-1.17 s
2026-10-08 00:32:13 +02:00
step23_ring_skip_capture_comparison.md
wip(standalone): GPU step 23 -- skip empty connector-ring dispatch; cached 1.07-1.17 s
2026-10-08 00:32:13 +02:00
step23_step22_control_capture_comparison.md
wip(standalone): GPU step 23 -- skip empty connector-ring dispatch; cached 1.07-1.17 s
2026-10-08 00:32:13 +02:00
step24_final384_capture_comparison.md
wip(standalone): GPU step 24 -- warmup policy off the critical path; first launch = CPU
2026-10-08 02:06:49 +02:00
step24_final_capture_comparison.md
wip(standalone): GPU step 24 -- warmup policy off the critical path; first launch = CPU
2026-10-08 02:06:49 +02:00
step25_final384_capture_comparison.md
wip(standalone): GPU step 25 -- shape profile, roughness endpoint reuse, export before CPU frame; first launch REGRESSED
2026-10-08 02:55:32 +02:00
step26_final384_capture_comparison.md
wip(standalone): GPU step 26 -- first launch fixed again; device-removed was our readback overrun
2026-10-08 03:22:38 +02:00
step27_final_skip_capture_comparison.md
wip(standalone): GPU step 27 -- slot-overflow fallback, -SkipCpuDetailBand (slower), -AffinityMask; full machine cached 0.93 s
2026-10-08 04:11:23 +02:00
step28_final_capture_comparison.md
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00
STEP4_REPORT.md
wip(standalone): GPU step 4 -- batched far generation, connector panels on GPU; far survival passes
2026-10-06 11:01:03 +02:00
STEP5_REPORT.md
wip(standalone): GPU step 5 -- far GPU work 646 -> 112 ms; hybrid whole view equal to CPU
2026-10-06 13:50:07 +02:00
STEP6_REPORT.md
wip(standalone): GPU step 6 -- near field on GPU, exact; cold start still slower than CPU
2026-10-06 20:03:10 +02:00
STEP7_RESUME_REPORT.md
wip(standalone): GPU step 7 -- viewer cold start 2.45 -> 1.49 s on CPU; GPU skirts missing
2026-10-06 22:39:40 +02:00
STEP9_REPORT.md
wip(standalone): GPU step 9 -- remote band closed by CPU fallback; CPU still wins whole view
2026-10-07 04:36:34 +02:00
STEP10_REPORT.md
wip(standalone): GPU step 10 -- first-submit stall bisected to cold shader use under 8 workers
2026-10-07 05:38:14 +02:00
STEP11_REPORT.md
wip(standalone): GPU step 11 -- per-shader cold cost measured; far EmitCS split, still CPU wins
2026-10-07 06:13:59 +02:00
STEP12_REPORT.md
wip(standalone): GPU step 12 -- far evaluator cold use 1704 -> 3 ms; near detail now the floor
2026-10-07 06:51:53 +02:00
STEP13_REPORT.md
wip(standalone): GPU step 13 -- near base/skirts/connectors cheaper to compile; warm-up still ~2.0 s
2026-10-07 07:41:31 +02:00
STEP14_REPORT.md
wip(standalone): GPU step 14 -- warm-up 2.04 -> 1.05 s; cold path now waits on streaming, not shaders
2026-10-07 08:03:00 +02:00
STEP15_REPORT.md
wip(standalone): GPU step 15 -- cold-start drain: hybrid whole view 3.2 -> 1.25-1.68 s
2026-10-07 08:47:36 +02:00
STEP16_REPORT.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
STEP17_REPORT.md
wip(standalone): GPU step 17 -- far skirt-depth fix (LOD3/4 back on GPU); adaptive start trades cached speed for first-launch
2026-10-07 20:36:40 +02:00
STEP18_REPORT.md
wip(standalone): GPU step 19 -- streamed GPU input export; RUNNERS.md
2026-10-07 22:09:29 +02:00
STEP19_REPORT.md
wip(standalone): GPU step 19 -- streamed GPU input export; RUNNERS.md
2026-10-07 22:09:29 +02:00
STEP20_REPORT.md
wip(standalone): GPU step 20 -- detail band batched by 8 columns; gap instrumented
2026-10-07 22:35:22 +02:00
STEP21_REPORT.md
wip(standalone): GPU step 21 -- GPU input export shares the 8 workers (3:1 priority); no wall gain
2026-10-07 23:20:42 +02:00
STEP22_REPORT.md
wip(standalone): GPU step 22 -- startup multi-batch pump, pool before handoff: cached 1.12-1.20 s
2026-10-08 00:00:06 +02:00
STEP23_REPORT.md
wip(standalone): GPU step 23 -- skip empty connector-ring dispatch; cached 1.07-1.17 s
2026-10-08 00:32:13 +02:00
STEP24_REPORT.md
wip(standalone): GPU step 24 -- warmup policy off the critical path; first launch = CPU
2026-10-08 02:06:49 +02:00
STEP25_REPORT.md
wip(standalone): GPU step 25 -- shape profile, roughness endpoint reuse, export before CPU frame; first launch REGRESSED
2026-10-08 02:55:32 +02:00
STEP26_REPORT.md
wip(standalone): GPU step 26 -- first launch fixed again; device-removed was our readback overrun
2026-10-08 03:22:38 +02:00
STEP27_REPORT.md
wip(standalone): GPU step 27 -- slot-overflow fallback, -SkipCpuDetailBand (slower), -AffinityMask; full machine cached 0.93 s
2026-10-08 04:11:23 +02:00
STEP28_REPORT.md
wip(standalone): GPU step 28 -- skirted near tiles on GPU (correct, but SLOWER); 16 shape workers
2026-10-08 17:07:40 +02:00