Fr0zka/VoxelForge
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
4b1108bf8654700c2d9a489d39f66c8a6e1c696a
VoxelForge/Standalone/Source/StreamBenchmarkPipeline.h
T
Fr0zkaandClaude Opus 5.5 4b1108bf86 wip(assembler): round L7 -- shared column cache across workers: whole view 16-19% faster, world unchanged 
SharedColumnCache.h: a world-owned, bounded 4-way second-tier cache of exact column values
(heightfield and floating islands), keyed by recipe/plan identity plus content hash, XY and island
id. Lock-free seqlock reads; set-local spin lock on misses. The FLazy thread-local tier stays
first. 83.6% of repeated column work was cross-worker; formula evaluations -60.9%.
W12 L1 view: Luna 2.165 -> 1.754 s; reviewer interleaved 2.059 -> 1.722 s. 8 m crossing new tiles
82 -> 52 ms. +66 MiB peak RSS. All hashes (whole view, per-LOD, exact queries, islands seed 0)
identical at W1/W8/W12, Clang off and GCC default. MSVC/UBT not compiled here.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CSFYgsdW8ejZSyiMf4qRSX
2026-10-19 21:46:15 +02:00

707 lines
41 KiB
C++
Raw Blame History

#pragma once
#include "OwnerWorldLoader.h"
#include "StreamingWorld.h"
#include "LinuxScreenSpaceGate.h"
#include <algorithm>
#include <array>
#include <chrono>
#include <cmath>
#include <cstdint>
#include <cstdio>
#include <filesystem>
#include <fstream>
#include <limits>
#include <thread>
#include <unordered_map>
#include <unordered_set>
#include <vector>
namespace StandaloneStreamBenchmark
{
struct FSceneExportWriter
{
std::ofstream File;
bool bGood=true;
std::uint32_t TileCount=0;
template<typename T> void Write(const T& Value)
{
File.write(reinterpret_cast<const char*>(&Value),sizeof(Value));
bGood=bGood&&static_cast<bool>(File);
}
bool Open(const std::filesystem::path& Path,std::uint32_t VoxelSizeMicrons,
const FVector& CameraVoxels,float YawDegrees,float PitchDegrees,
std::uint32_t ExpectedTiles)
{
std::error_code Error;
if(!Path.parent_path().empty())
std::filesystem::create_directories(Path.parent_path(),Error);
if(Error)return false;
File.open(Path,std::ios::binary|std::ios::trunc);
if(!File)return false;
constexpr char Magic[8]={'V','F','M','S','H','0','0','1'};
File.write(Magic,sizeof(Magic));
const std::uint32_t Version=1;
Write(Version);Write(VoxelSizeMicrons);Write(ExpectedTiles);
const double MicronsToMeters=static_cast<double>(VoxelSizeMicrons)/1000000.0;
const double Camera[3]={CameraVoxels.X*MicronsToMeters,
CameraVoxels.Y*MicronsToMeters,CameraVoxels.Z*MicronsToMeters};
for(const double Value:Camera)Write(Value);
const double Yaw=YawDegrees,Pitch=PitchDegrees;
Write(Yaw);Write(Pitch);
return bGood;
}
void Add(const FStandaloneCpuTile& Tile)
{
const std::int32_t Key[4]={Tile.Key.Coord.X,Tile.Key.Coord.Y,
Tile.Key.Coord.Z,Tile.Key.Level};
for(const auto Value:Key)Write(Value);
const std::uint32_t SurfaceVertexCount=Tile.SurfaceVertexCount;
const std::uint32_t SkirtTriangleCount=Tile.SkirtTriangleCount;
const std::uint32_t VertexCount=static_cast<std::uint32_t>(Tile.Vertices.size());
const std::uint32_t IndexCount=static_cast<std::uint32_t>(Tile.Indices.size());
const std::uint32_t MaskCount=static_cast<std::uint32_t>(Tile.TriangleSkirtFaceMasks.size());
Write(SurfaceVertexCount);Write(SkirtTriangleCount);
Write(VertexCount);Write(IndexCount);Write(MaskCount);
for(const FStandaloneCpuVertex& Vertex:Tile.Vertices)
for(const float Value:Vertex.Position)Write(Value);
for(const FStandaloneCpuVertex& Vertex:Tile.Vertices)
for(const float Value:Vertex.Normal)Write(Value);
if(IndexCount)
File.write(reinterpret_cast<const char*>(Tile.Indices.data()),
static_cast<std::streamsize>(sizeof(std::uint32_t)*Tile.Indices.size()));
if(MaskCount)
File.write(reinterpret_cast<const char*>(Tile.TriangleSkirtFaceMasks.data()),
static_cast<std::streamsize>(Tile.TriangleSkirtFaceMasks.size()));
bGood=bGood&&static_cast<bool>(File);
++TileCount;
}
bool Close(std::uint32_t ExpectedTiles)
{
File.flush();
const bool bComplete=bGood&&static_cast<bool>(File)&&TileCount==ExpectedTiles;
File.close();
return bComplete;
}
};
inline void HashWord(uint64& Hash,uint64 Word,int32 Bytes=8)
{
for(int32 I=0;I<Bytes;++I)
{
Hash^=static_cast<uint8>(Word>>(I*8));
Hash*=1099511628211ull;
}
}
struct FHashedTile
{
FVoxelTileKey Key;
uint64 Hash=14695981039346656037ull;
};
inline bool TileKeyLessForHash(const FVoxelTileKey& A,const FVoxelTileKey& B)
{
if(A.Level!=B.Level)return A.Level<B.Level;
if(A.Coord.Z!=B.Coord.Z)return A.Coord.Z<B.Coord.Z;
if(A.Coord.Y!=B.Coord.Y)return A.Coord.Y<B.Coord.Y;
return A.Coord.X<B.Coord.X;
}
inline uint64 HashTileRows(std::vector<FHashedTile>& Rows)
{
std::sort(Rows.begin(),Rows.end(),[](const FHashedTile& A,const FHashedTile& B)
{return TileKeyLessForHash(A.Key,B.Key);});
uint64 Hash=14695981039346656037ull;
for(const FHashedTile& Row:Rows)
{
HashWord(Hash,static_cast<uint32>(Row.Key.Level),4);
HashWord(Hash,static_cast<uint32>(Row.Key.Coord.X),4);
HashWord(Hash,static_cast<uint32>(Row.Key.Coord.Y),4);
HashWord(Hash,static_cast<uint32>(Row.Key.Coord.Z),4);
HashWord(Hash,Row.Hash);
}
return Hash;
}
inline int RunPipelinedStreamingBenchmark(std::shared_ptr<FStandaloneOwnerWorld> OwnerWorld,
const VoxelForgeCore::ResolvedWorldSnapshot& CoreSnapshot,
const FVector& CameraPositionVoxels,int32 WorkerCount,int32 MaxLevel,
VoxelForgeCore::DetailVariant DetailVariant,const VoxelForgeCore::JobScheduler& Scheduler,
bool bHashDensity,bool bSkipUvAndColors,bool bSkipMaterialGroups,bool bSkipUV,bool bSkipColors,
std::chrono::steady_clock::time_point ColdStart,bool bProfileColumnOverlap,
int32 FarHorizonLOD8InteriorStepVoxels,bool bReportMeshCoverage=false,
const std::filesystem::path& MeshExportPath={},float CameraYawDegrees=0.0f,
float CameraPitchDegrees=0.0f,bool bScreenSpaceGate=false,
bool bUseSharedColumnCache=true,FVector CameraStepVoxels=FVector::ZeroVector)
{
using Clock=std::chrono::steady_clock;
FStandaloneOwnerWorld& World=*OwnerWorld;
const VoxelForgeCore::WorldAssemblyManifest* Assembly=CoreSnapshot.worldAssembly();
std::array<double,9> ClosestDisplayDistances{};
std::array<std::int64_t,9> PixelBudgetsSubvoxels{};
if(bScreenSpaceGate)
{
ClosestDisplayDistances=StandaloneScreenSpaceGate::ClosestDisplayedDistancesMeters(
World,CameraPositionVoxels,World.CoreVoxelSizeMicrons);
const double VoxelMeters=static_cast<double>(World.CoreVoxelSizeMicrons)/1000000.0;
for(int32 Lod=3;Lod<=World.MaxClipLevel&&Lod<9;++Lod)
PixelBudgetsSubvoxels[static_cast<std::size_t>(Lod)]=static_cast<std::int64_t>(std::floor(
ClosestDisplayDistances[static_cast<std::size_t>(Lod)]
*StandaloneScreenSpaceGate::OnePixelRadians/VoxelMeters
*VoxelForgeCore::WorldAssemblySubvoxelsPerVoxel));
}
const bool bRoundTunnels=World.Generator->bStandaloneResearchRoundTunnels;
const bool bEnhancedTunnelWalls=World.Generator->bStandaloneResearchEnhancedWalls;
VoxelForgeCore::World Shape(CoreSnapshot);
FStandaloneDesiredTileSet Desired=StandaloneWorldMesher::BuildDesiredSet(World,CameraPositionVoxels);
const auto CoreTiles=Shape.desiredTiles(
{CameraPositionVoxels.X,CameraPositionVoxels.Y,CameraPositionVoxels.Z});
Desired.Tiles.clear();Desired.Tiles.reserve(CoreTiles.size());
for(const auto& Key:CoreTiles)
Desired.Tiles.emplace_back(FIntVector(Key.coordinate.x,Key.coordinate.y,Key.coordinate.z),Key.lod);
if(MaxLevel>=0)
Desired.Tiles.erase(std::remove_if(Desired.Tiles.begin(),Desired.Tiles.end(),
[&](const FVoxelTileKey& Tile){return Tile.Level>MaxLevel;}),Desired.Tiles.end());
const bool bHasCameraStep=std::abs(CameraStepVoxels.X)+std::abs(CameraStepVoxels.Y)
+std::abs(CameraStepVoxels.Z)>1.0e-6f;
FStandaloneDesiredTileSet MovedDesired;
if(bHasCameraStep)
{
const FVector MovedCamera=CameraPositionVoxels+CameraStepVoxels;
MovedDesired=StandaloneWorldMesher::BuildDesiredSet(World,MovedCamera);
const auto MovedCoreTiles=Shape.desiredTiles({MovedCamera.X,MovedCamera.Y,MovedCamera.Z});
MovedDesired.Tiles.clear();MovedDesired.Tiles.reserve(MovedCoreTiles.size());
for(const auto& Key:MovedCoreTiles)
MovedDesired.Tiles.emplace_back(FIntVector(Key.coordinate.x,Key.coordinate.y,Key.coordinate.z),Key.lod);
if(MaxLevel>=0)
MovedDesired.Tiles.erase(std::remove_if(MovedDesired.Tiles.begin(),MovedDesired.Tiles.end(),
[&](const FVoxelTileKey& Tile){return Tile.Level>MaxLevel;}),MovedDesired.Tiles.end());
}
if(!Desired.Tiles.empty())
{
FIntVector volumeMin(INT32_MAX,INT32_MAX,INT32_MAX);
FIntVector volumeMax(INT32_MIN,INT32_MIN,INT32_MIN);
int32 minimumLod=INT32_MAX,maximumLod=INT32_MIN;
for(const FVoxelTileKey& tile:Desired.Tiles)
{
const FIntVector origin=tile.OriginVoxels();
const int32 extent=tile.ExtentVoxels();
volumeMin.X=std::min(volumeMin.X,origin.X);
volumeMin.Y=std::min(volumeMin.Y,origin.Y);
volumeMin.Z=std::min(volumeMin.Z,origin.Z);
volumeMax.X=std::max(volumeMax.X,origin.X+extent);
volumeMax.Y=std::max(volumeMax.Y,origin.Y+extent);
volumeMax.Z=std::max(volumeMax.Z,origin.Z+extent);
minimumLod=std::min(minimumLod,tile.Level);
maximumLod=std::max(maximumLod,tile.Level);
}
std::printf("[PipelineVolume] tiles=%zu lod=%d..%d min_voxel=(%d,%d,%d) max_voxel=(%d,%d,%d)\n",
Desired.Tiles.size(),minimumLod,maximumLod,volumeMin.X,volumeMin.Y,volumeMin.Z,
volumeMax.X,volumeMax.Y,volumeMax.Z);
}
FIntVector FineMin(INT32_MAX,INT32_MAX,INT32_MAX),FineMax(INT32_MIN,INT32_MIN,INT32_MIN);
FIntVector ClipMin(INT32_MAX,INT32_MAX,INT32_MAX),ClipMax(INT32_MIN,INT32_MIN,INT32_MIN);
const auto AccumulateBounds=[&](const std::vector<FVoxelTileKey>& Tiles)
{
for(const FVoxelTileKey& Tile:Tiles)
{
const FIntVector Origin=Tile.OriginVoxels();const int32 Extent=Tile.ExtentVoxels();
ClipMin.X=std::min(ClipMin.X,Origin.X);ClipMin.Y=std::min(ClipMin.Y,Origin.Y);ClipMin.Z=std::min(ClipMin.Z,Origin.Z);
ClipMax.X=std::max(ClipMax.X,Origin.X+Extent);ClipMax.Y=std::max(ClipMax.Y,Origin.Y+Extent);ClipMax.Z=std::max(ClipMax.Z,Origin.Z+Extent);
if(Tile.Level<World.FullResClipLevels)
{
FineMin.X=std::min(FineMin.X,Origin.X);FineMin.Y=std::min(FineMin.Y,Origin.Y);FineMin.Z=std::min(FineMin.Z,Origin.Z);
FineMax.X=std::max(FineMax.X,Origin.X+Extent);FineMax.Y=std::max(FineMax.Y,Origin.Y+Extent);FineMax.Z=std::max(FineMax.Z,Origin.Z+Extent);
}
}
};
AccumulateBounds(Desired.Tiles);
if(bHasCameraStep)AccumulateBounds(MovedDesired.Tiles);
if(FineMin.X==INT32_MAX||ClipMin.X==INT32_MAX)
{std::fprintf(stderr,"[StreamPipeline] no desired detail or clipmap tiles.\n");return 7;}
const FIntVector CenterChunk=Desired.CenterChunk;
std::unordered_set<FVoxelTileKey,FStandaloneTileKeyHash> Neighborhood;
Neighborhood.reserve(32);
for(const FVoxelTileKey& Tile:Desired.Tiles)
if(Tile.Level==0&&std::abs(Tile.Coord.X-CenterChunk.X)<=1
&&std::abs(Tile.Coord.Y-CenterChunk.Y)<=1&&std::abs(Tile.Coord.Z-CenterChunk.Z)<=1)
Neighborhood.insert(Tile);
if(Neighborhood.size()!=27)
{std::fprintf(stderr,"[StreamPipeline] expected 27 player-neighborhood tiles, found %zu.\n",Neighborhood.size());return 7;}
// The initial desired set now owns all clipmap metadata needed by this Core-only run.
// Release the separate compatibility-world generation graph before compiling Core.
World.Generator.reset();
World.Manager.reset();
World.DefinitionStorage.clear();
World.BiomeStorage.clear();
World.OperationStorage.clear();
World.DensitySamples.clear();
std::string BuildError;
const bool bIncludeFarWorld=MaxLevel<0||MaxLevel>=World.FullResClipLevels;
const auto ShapeBuildStart=Clock::now();
VoxelForgeCore::ShapeBuildRequest ShapeRequest;
ShapeRequest.minVoxel={FineMin.X,FineMin.Y,FineMin.Z};
ShapeRequest.maxVoxel={FineMax.X,FineMax.Y,FineMax.Z};
ShapeRequest.clipMinVoxel={ClipMin.X,ClipMin.Y,ClipMin.Z};
ShapeRequest.clipMaxVoxel={ClipMax.X,ClipMax.Y,ClipMax.Z};
ShapeRequest.workerLimit=static_cast<uint32>(std::clamp(WorkerCount,1,16));
ShapeRequest.detail=DetailVariant;ShapeRequest.includeFarWorld=bIncludeFarWorld;
ShapeRequest.roundTunnels=bRoundTunnels;
ShapeRequest.enhancedTunnelWalls=bEnhancedTunnelWalls;
if(!Shape.compile(ShapeRequest,Scheduler,BuildError))
{std::fprintf(stderr,"[StreamPipeline] full shape build failed: %s\n",BuildError.c_str());return 2;}
const auto ShapeReady=Clock::now();
const double ShapeSeconds=std::chrono::duration<double>(ShapeReady-ShapeBuildStart).count();
FStandaloneStreamingGenerator Streamer(World,WorkerCount,bHashDensity,nullptr,MaxLevel,
bSkipUvAndColors,bSkipMaterialGroups,bSkipUV,bSkipColors,&Shape,bProfileColumnOverlap,
FarHorizonLOD8InteriorStepVoxels,bUseSharedColumnCache);
if(bScreenSpaceGate)Streamer.SetMesherStageProfilingEnabled(true);
const auto StreamStart=Clock::now();
const FStandaloneStreamingUpdate Update=Streamer.StartWithDesired(Desired);
if(Update.DesiredTiles!=static_cast<int32>(Desired.Tiles.size()))
{
Streamer.Stop();
std::fprintf(stderr,"[StreamPipeline] expected %zu tiles but queued %d.\n",
Desired.Tiles.size(),Update.DesiredTiles);return 7;
}
FSceneExportWriter SceneExport;
const bool bExportRequested=!MeshExportPath.empty();
bool bSceneExportPassing=true;
if(bExportRequested)
{
bSceneExportPassing=SceneExport.Open(MeshExportPath,World.CoreVoxelSizeMicrons,
CameraPositionVoxels,CameraYawDegrees,CameraPitchDegrees,
static_cast<std::uint32_t>(Update.DesiredTiles));
if(!bSceneExportPassing)
{
Streamer.Stop();
std::fprintf(stderr,"[MeshSceneExport] FAIL: could not open %s\n",
MeshExportPath.string().c_str());
return 8;
}
}
std::vector<FHashedTile> DensityRows,MeshRows,PayloadRows,TopologyRows;
int32 Finished=0,NeighborhoodFinished=0,FullDetailFinished=0;
int32 FullDetailExpected=0;
uint64 WholeNonEmptyMeshTiles=0,WholeTriangleCount=0;
struct FColumnXY
{
std::int64_t X=0,Y=0;
bool operator==(const FColumnXY&)const=default;
};
struct FColumnXYHash
{
std::size_t operator()(const FColumnXY& Value)const noexcept
{
const std::size_t X=std::hash<std::int64_t>{}(Value.X);
const std::size_t Y=std::hash<std::int64_t>{}(Value.Y);
return X^(Y+static_cast<std::size_t>(0x9E3779B9u)+(X<<6)+(X>>2));
}
};
struct FColumnEvaluationEvent
{
FColumnXY XY;
FVoxelTileKey Tile;
std::uint64_t WorkerHash=0;
std::uint64_t Sequence=0;
bool bOutsideTileFootprint=false;
};
std::array<std::vector<FColumnEvaluationEvent>,9> ColumnEventsByLOD;
std::array<std::uint64_t,9> ColumnEvaluationEventsByLOD{},UniqueColumnCoordinatesByLOD{};
std::array<std::uint64_t,9> ColumnDuplicateEventsByLOD{},SameWorkerEvictedRepeatsByLOD{};
std::array<std::uint64_t,9> DifferentWorkerRepeatsByLOD{},HaloNormalRepeatsByLOD{};
std::array<std::uint64_t,9> HaloNormalEvaluationsByLOD{},NeighbourDuplicateEventsByLOD{};
std::array<std::uint64_t,9> CrossWorkerNeighbourDuplicateEventsByLOD{};
std::array<std::uint64_t,9> ColumnCaptureTilesByLOD{},ExpectedTilesByLOD{};
std::array<bool,9> ColumnCaptureCompleteByLOD{};
ColumnCaptureCompleteByLOD.fill(true);
std::array<double,9> WorkerSecondsByLOD{};
std::array<double,9> ColumnEvalMsByLOD{},MarchingExclusiveMsByLOD{};
std::array<double,9> GroundRefineExclusiveMsByLOD{},CapRefineExclusiveMsByLOD{};
std::array<double,9> MeshAssemblyMsByLOD{};
std::array<double,9> SkirtMsByLOD{};
std::array<std::uint64_t,9> SkirtTrianglesByLOD{};
std::array<std::uint64_t,9> TileCountByLOD{};
for(const FVoxelTileKey& Tile:Desired.Tiles)
if(Tile.Level>=0&&Tile.Level<static_cast<int32>(ExpectedTilesByLOD.size()))
++ExpectedTilesByLOD[static_cast<std::size_t>(Tile.Level)];
const auto AreSameLODNeighbours=[](const FVoxelTileKey& A,const FVoxelTileKey& B)
{
if(A.Level!=B.Level)return false;
const std::int64_t DX=std::abs(static_cast<std::int64_t>(A.Coord.X)-B.Coord.X);
const std::int64_t DY=std::abs(static_cast<std::int64_t>(A.Coord.Y)-B.Coord.Y);
const std::int64_t DZ=std::abs(static_cast<std::int64_t>(A.Coord.Z)-B.Coord.Z);
return std::max({DX,DY,DZ})<=1&&(DX+DY+DZ)>0;
};
uint64 NativeSolidBoxes=0,NativeAirBoxes=0,NativeMixedBoxes=0,NativeUnknownBoxes=0;
uint64 NativeFieldCalls=0,NativeColumnHits=0,NativeColumnMisses=0,NativeMeshingSkips=0;
double NativeClassificationMilliseconds=0.0,NativeColumnComputeMilliseconds=0.0;
double NativeMeshingMilliseconds=0.0;
for(const FVoxelTileKey& Tile:Desired.Tiles)
if(Tile.Level<World.FullResClipLevels)++FullDetailExpected;
std::array<std::optional<FStandaloneCpuTile>,9> GateSampleTiles{};
std::array<double,9> GateSampleDistances{};
GateSampleDistances.fill(std::numeric_limits<double>::infinity());
Clock::time_point NeighborhoodReady{},FullDetailReady{},WholeViewReady{};
bool bNeighborhoodReported=false,bFullDetailReported=false;
while(Finished<Update.DesiredTiles)
{
std::vector<FStandaloneCpuTile> Ready=Streamer.TakeReady();
if(Ready.empty()){std::this_thread::sleep_for(std::chrono::milliseconds(1));continue;}
for(const FStandaloneCpuTile& Tile:Ready)
{
if(bExportRequested)SceneExport.Add(Tile);
++Finished;
if(bReportMeshCoverage)
{
const uint64 TileTriangleCount=static_cast<uint64>(Tile.Indices.size()/3);
WholeTriangleCount+=TileTriangleCount;
if(TileTriangleCount>0)++WholeNonEmptyMeshTiles;
}
if(Tile.Key.Level>=0&&Tile.Key.Level<static_cast<int32>(TileCountByLOD.size()))
{
const std::size_t LOD=static_cast<std::size_t>(Tile.Key.Level);
++TileCountByLOD[LOD];
WorkerSecondsByLOD[LOD]+=Tile.WorkerSeconds;
ColumnEvalMsByLOD[LOD]+=Tile.CoreDiagnostics.columnEvaluationExclusiveMilliseconds;
MarchingExclusiveMsByLOD[LOD]+=Tile.CoreDiagnostics.marchingCubesExclusiveMilliseconds;
GroundRefineExclusiveMsByLOD[LOD]+=Tile.CoreDiagnostics.groundRefinementExclusiveMilliseconds;
CapRefineExclusiveMsByLOD[LOD]+=Tile.CoreDiagnostics.capRefinementExclusiveMilliseconds;
MeshAssemblyMsByLOD[LOD]+=Tile.CoreDiagnostics.meshAssemblyMilliseconds;
SkirtMsByLOD[LOD]+=Tile.CoreDiagnostics.marchingCubesSkirtMilliseconds;
SkirtTrianglesByLOD[LOD]+=Tile.SkirtTriangleCount;
if(bScreenSpaceGate&&Assembly&&Tile.Key.Level>=3&&Tile.Key.Level<9
&&!Tile.Vertices.empty()
&&StandaloneScreenSpaceGate::TileHasNativeSurface(Tile.Key,*Assembly,
static_cast<double>(World.CoreVoxelSizeMicrons)/1000000.0))
{
const double Distance=StandaloneScreenSpaceGate::TileAabbDistanceMeters(
Tile.Key,CameraPositionVoxels,
static_cast<double>(World.CoreVoxelSizeMicrons)/1000000.0);
const bool bCloser=Distance<GateSampleDistances[LOD]-1e-9;
const bool bTied=std::abs(Distance-GateSampleDistances[LOD])<=1e-9;
const bool bKeyLess=!GateSampleTiles[LOD]
||TileKeyLessForHash(Tile.Key,GateSampleTiles[LOD]->Key);
if(bCloser||(bTied&&bKeyLess))
{GateSampleDistances[LOD]=Distance;GateSampleTiles[LOD]=Tile;}
}
}
if(bProfileColumnOverlap&&Tile.Key.Level>=0
&&Tile.Key.Level<static_cast<int32>(ColumnCaptureTilesByLOD.size()))
{
const std::size_t LOD=static_cast<std::size_t>(Tile.Key.Level);
++ColumnCaptureTilesByLOD[LOD];
ColumnCaptureCompleteByLOD[LOD]=ColumnCaptureCompleteByLOD[LOD]
&&Tile.CoreDiagnostics.columnCoordinateCaptureComplete;
auto& Events=ColumnEventsByLOD[LOD];
for(const auto& Coordinate:Tile.CoreDiagnostics.evaluatedColumnCoordinates)
{
Events.push_back({{Coordinate.x,Coordinate.y},Tile.Key,
Tile.CoreDiagnostics.columnEvaluationWorkerHash,Coordinate.sequence,
Coordinate.outsideTileFootprint});
}
}
WholeViewReady=std::max(WholeViewReady,Tile.CompletedAt);
if(Neighborhood.contains(Tile.Key))
{
++NeighborhoodFinished;
NeighborhoodReady=std::max(NeighborhoodReady,Tile.CompletedAt);
}
if(Tile.Key.Level<World.FullResClipLevels)
{
++FullDetailFinished;
FullDetailReady=std::max(FullDetailReady,Tile.CompletedAt);
NativeSolidBoxes+=Tile.CoreDiagnostics.solidBoxes;
NativeAirBoxes+=Tile.CoreDiagnostics.airBoxes;
NativeMixedBoxes+=Tile.CoreDiagnostics.mixedBoxes;
NativeUnknownBoxes+=Tile.CoreDiagnostics.unknownBoxes;
NativeFieldCalls+=Tile.CoreDiagnostics.fieldCalls;
NativeColumnHits+=Tile.CoreDiagnostics.columnCacheHits;
NativeColumnMisses+=Tile.CoreDiagnostics.columnCacheMisses;
NativeMeshingSkips+=Tile.CoreDiagnostics.meshingSkipped?1u:0u;
NativeClassificationMilliseconds+=Tile.CoreDiagnostics.classificationMilliseconds;
NativeColumnComputeMilliseconds+=Tile.CoreDiagnostics.columnComputeMilliseconds;
NativeMeshingMilliseconds+=Tile.CoreDiagnostics.meshingMilliseconds;
}
if(bHashDensity)
{
if(Tile.bHasDensityHash)DensityRows.push_back({Tile.Key,Tile.DensityHash});
if(Tile.bHasMeshHash)MeshRows.push_back({Tile.Key,Tile.MeshHash});
if(Tile.bHasPayloadHash)PayloadRows.push_back({Tile.Key,Tile.PayloadHash});
if(Tile.bHasTopologyHash)TopologyRows.push_back({Tile.Key,Tile.TopologyHash});
}
}
if(!bNeighborhoodReported&&NeighborhoodFinished==static_cast<int32>(Neighborhood.size()))
{
std::printf("[PipelineMilestone] neighborhood_ready=%.3fs generated=%d/27\n",
std::chrono::duration<double>(NeighborhoodReady-ColdStart).count(),NeighborhoodFinished);
bNeighborhoodReported=true;
}
if(!bFullDetailReported&&FullDetailFinished==FullDetailExpected)
{
std::printf("[PipelineMilestone] full_detail_ready=%.3fs generated=%d/%d\n",
std::chrono::duration<double>(FullDetailReady-ColdStart).count(),
FullDetailFinished,FullDetailExpected);
bFullDetailReported=true;
}
}
if(bHasCameraStep)
{
const auto MovingStart=Clock::now();
const auto MovingUpdate=Streamer.UpdateDesiredWithSet(MovedDesired,Desired.Tiles,1);
std::array<std::uint64_t,9> MovingTilesByLOD{};
std::array<double,9> MovingWorkerSecondsByLOD{};
int32 MovingFinished=0;
while(MovingFinished<MovingUpdate.NewJobs)
{
std::vector<FStandaloneCpuTile> Ready=Streamer.TakeReady();
if(Ready.empty())
{std::this_thread::sleep_for(std::chrono::milliseconds(1));continue;}
for(const FStandaloneCpuTile& Tile:Ready)
{
++MovingFinished;
if(Tile.Key.Level>=0&&Tile.Key.Level<static_cast<int32>(MovingTilesByLOD.size()))
{
const std::size_t LOD=static_cast<std::size_t>(Tile.Key.Level);
++MovingTilesByLOD[LOD];
MovingWorkerSecondsByLOD[LOD]+=Tile.WorkerSeconds;
}
}
}
const double MovingWallSeconds=std::chrono::duration<double>(Clock::now()-MovingStart).count();
const double StepDistanceMeters=std::sqrt(
static_cast<double>(CameraStepVoxels.X)*CameraStepVoxels.X
+static_cast<double>(CameraStepVoxels.Y)*CameraStepVoxels.Y
+static_cast<double>(CameraStepVoxels.Z)*CameraStepVoxels.Z)
*static_cast<double>(World.CoreVoxelSizeMicrons)/1000000.0;
std::printf("[MovingCameraStep] step_voxels=(%.3f,%.3f,%.3f) distance_m=%.3f desired_tiles=%zu new_tiles=%d completed=%d wall_s=%.6f shared_column_cache=%s\n",
CameraStepVoxels.X,CameraStepVoxels.Y,CameraStepVoxels.Z,StepDistanceMeters,
MovedDesired.Tiles.size(),MovingUpdate.NewJobs,MovingFinished,MovingWallSeconds,
bUseSharedColumnCache?"on":"off");
for(std::size_t LOD=0;LOD<MovingTilesByLOD.size();++LOD)
if(MovingTilesByLOD[LOD]!=0)
std::printf("[MovingCameraStepLOD] lod=%zu tiles=%llu worker_sum_s=%.6f\n",
LOD,static_cast<unsigned long long>(MovingTilesByLOD[LOD]),
MovingWorkerSecondsByLOD[LOD]);
}
Streamer.Stop();
bool bScreenSpaceGatePassing=true;
if(bScreenSpaceGate)
{
if(Assembly)
bScreenSpaceGatePassing=StandaloneScreenSpaceGate::Run(GateSampleTiles,
Streamer,*Assembly,ClosestDisplayDistances,World.CoreVoxelSizeMicrons,
World.MaxClipLevel,PixelBudgetsSubvoxels,CameraYawDegrees,
CameraPitchDegrees);
else
{std::fprintf(stderr,"[ScreenSpaceGateResult] no-world-assembly status=REJECT\n");bScreenSpaceGatePassing=false;}
}
if(bExportRequested)
{
bSceneExportPassing=SceneExport.Close(static_cast<std::uint32_t>(Update.DesiredTiles));
std::error_code FileSizeError;
const std::uintmax_t ExportBytes=std::filesystem::file_size(MeshExportPath,FileSizeError);
std::printf("[MeshSceneExport] tiles=%u expected=%d bytes=%llu status=%s path=%s\n",
SceneExport.TileCount,Update.DesiredTiles,
static_cast<unsigned long long>(FileSizeError?0:ExportBytes),
bSceneExportPassing?"PASS":"FAIL",MeshExportPath.string().c_str());
bSceneExportPassing=bSceneExportPassing&&!FileSizeError;
}
if(bProfileColumnOverlap)
{
struct FColumnPriorUse
{
FVoxelTileKey Tile;
std::uint64_t WorkerHash=0;
};
for(std::size_t LOD=0;LOD<ColumnEventsByLOD.size();++LOD)
{
auto& Events=ColumnEventsByLOD[LOD];
std::sort(Events.begin(),Events.end(),[](const FColumnEvaluationEvent& A,
const FColumnEvaluationEvent& B){return A.Sequence<B.Sequence;});
std::unordered_map<FColumnXY,std::vector<FColumnPriorUse>,FColumnXYHash> UsesByXY;
UsesByXY.reserve(Events.size());
for(const FColumnEvaluationEvent& Event:Events)
{
++ColumnEvaluationEventsByLOD[LOD];
if(Event.bOutsideTileFootprint)++HaloNormalEvaluationsByLOD[LOD];
auto [Found,Inserted]=UsesByXY.try_emplace(Event.XY);
auto& PriorUses=Found->second;
if(Inserted)
{
++UniqueColumnCoordinatesByLOD[LOD];
PriorUses.push_back({Event.Tile,Event.WorkerHash});
continue;
}
++ColumnDuplicateEventsByLOD[LOD];
if(Event.bOutsideTileFootprint)
++HaloNormalRepeatsByLOD[LOD];
else
{
const bool bSameWorkerSeen=std::any_of(PriorUses.begin(),PriorUses.end(),
[&](const FColumnPriorUse& Prior)
{return Prior.WorkerHash==Event.WorkerHash;});
if(bSameWorkerSeen)++SameWorkerEvictedRepeatsByLOD[LOD];
else ++DifferentWorkerRepeatsByLOD[LOD];
}
bool bHasNeighbourDuplicate=false,bHasCrossWorkerNeighbourDuplicate=false;
for(const FColumnPriorUse& Prior:PriorUses)
{
if(!AreSameLODNeighbours(Prior.Tile,Event.Tile))continue;
bHasNeighbourDuplicate=true;
bHasCrossWorkerNeighbourDuplicate=bHasCrossWorkerNeighbourDuplicate
||Prior.WorkerHash!=Event.WorkerHash;
}
if(bHasNeighbourDuplicate)++NeighbourDuplicateEventsByLOD[LOD];
if(bHasCrossWorkerNeighbourDuplicate)
++CrossWorkerNeighbourDuplicateEventsByLOD[LOD];
PriorUses.push_back({Event.Tile,Event.WorkerHash});
}
}
}
bool bColumnOverlapCapturePassing=true;
if(bProfileColumnOverlap)
{
for(std::size_t LOD=0;LOD<ColumnCaptureTilesByLOD.size();++LOD)
{
const bool bLODComplete=ColumnCaptureCompleteByLOD[LOD]
&&ColumnCaptureTilesByLOD[LOD]==ExpectedTilesByLOD[LOD];
bColumnOverlapCapturePassing=bColumnOverlapCapturePassing&&bLODComplete;
std::printf("[WorldAssemblerLODColumnOverlap] lod=%zu tiles=%llu expected_tiles=%llu total_column_evaluations=%llu unique_xy_columns=%llu repeated_column_evaluations=%llu same_worker_evicted_repeats=%llu different_worker_repeats=%llu halo_normal_repeats=%llu halo_normal_evaluations=%llu neighbouring_duplicate_evaluations=%llu cross_worker_neighbouring_duplicates=%llu capture_complete=%d\n",
LOD,static_cast<unsigned long long>(ColumnCaptureTilesByLOD[LOD]),
static_cast<unsigned long long>(ExpectedTilesByLOD[LOD]),
static_cast<unsigned long long>(ColumnEvaluationEventsByLOD[LOD]),
static_cast<unsigned long long>(UniqueColumnCoordinatesByLOD[LOD]),
static_cast<unsigned long long>(ColumnDuplicateEventsByLOD[LOD]),
static_cast<unsigned long long>(SameWorkerEvictedRepeatsByLOD[LOD]),
static_cast<unsigned long long>(DifferentWorkerRepeatsByLOD[LOD]),
static_cast<unsigned long long>(HaloNormalRepeatsByLOD[LOD]),
static_cast<unsigned long long>(HaloNormalEvaluationsByLOD[LOD]),
static_cast<unsigned long long>(NeighbourDuplicateEventsByLOD[LOD]),
static_cast<unsigned long long>(CrossWorkerNeighbourDuplicateEventsByLOD[LOD]),
bLODComplete?1:0);
}
}
for(std::size_t LOD=0;LOD<TileCountByLOD.size();++LOD)
{
std::printf("[WorldAssemblerLODWorker] lod=%zu tiles=%llu worker_sum_s=%.6f column_eval_exclusive_ms=%.3f mc_exclusive_ms=%.3f ground_refine_exclusive_ms=%.3f cap_refine_exclusive_ms=%.3f skirt_ms=%.3f skirt_triangles=%llu mesh_assembly_ms=%.3f\n",
LOD,static_cast<unsigned long long>(TileCountByLOD[LOD]),WorkerSecondsByLOD[LOD],
ColumnEvalMsByLOD[LOD],MarchingExclusiveMsByLOD[LOD],
GroundRefineExclusiveMsByLOD[LOD],CapRefineExclusiveMsByLOD[LOD],
SkirtMsByLOD[LOD],static_cast<unsigned long long>(SkirtTrianglesByLOD[LOD]),
MeshAssemblyMsByLOD[LOD]);
}
const double NeighborhoodSeconds=std::chrono::duration<double>(NeighborhoodReady-ColdStart).count();
const double NeighborhoodMeshSeconds=std::chrono::duration<double>(NeighborhoodReady-ShapeReady).count();
const double FullDetailSeconds=std::chrono::duration<double>(FullDetailReady-ColdStart).count();
const double WholeViewSeconds=std::chrono::duration<double>(WholeViewReady-ColdStart).count();
const double StreamSeconds=std::chrono::duration<double>(WholeViewReady-StreamStart).count();
std::printf("[StreamPipeline] workers=%d shape=%.3fs neighborhood_mesh=%.3fs neighborhood_ready=%.3fs full_detail_ready=%.3fs whole_view_ready=%.3fs neighborhood_tiles=%d/27 full_detail_tiles=%d/%d whole_tiles=%d/%zu stream_wall=%.3fs cache_hits=0 full_payload=1\n",
WorkerCount,ShapeSeconds,NeighborhoodMeshSeconds,NeighborhoodSeconds,
FullDetailSeconds,WholeViewSeconds,NeighborhoodFinished,
FullDetailFinished,FullDetailExpected,Finished,Desired.Tiles.size(),StreamSeconds);
if(bReportMeshCoverage)
std::printf("[PipelineMeshCoverage] tiles=%d nonempty_mesh_tiles=%llu indexed_triangles=%llu\n",
Finished,static_cast<unsigned long long>(WholeNonEmptyMeshTiles),
static_cast<unsigned long long>(WholeTriangleCount));
std::printf("[WorldAssemblerTilePipeline] full_detail_tiles=%d solid_boxes=%llu air_boxes=%llu mixed_boxes=%llu unknown_boxes=%llu whole_tile_skips=%llu field_calls=%llu column_cache_hits=%llu column_cache_misses=%llu column_compute_worker_ms=%.3f classification_worker_ms=%.3f meshing_worker_ms=%.3f\n",
FullDetailExpected,static_cast<unsigned long long>(NativeSolidBoxes),
static_cast<unsigned long long>(NativeAirBoxes),
static_cast<unsigned long long>(NativeMixedBoxes),
static_cast<unsigned long long>(NativeUnknownBoxes),
static_cast<unsigned long long>(NativeMeshingSkips),
static_cast<unsigned long long>(NativeFieldCalls),
static_cast<unsigned long long>(NativeColumnHits),
static_cast<unsigned long long>(NativeColumnMisses),
NativeColumnComputeMilliseconds,NativeClassificationMilliseconds,
NativeMeshingMilliseconds);
bool bHashesPassing=true;
if(bHashDensity)
{
bHashesPassing=DensityRows.size()==Desired.Tiles.size()
&&MeshRows.size()==Desired.Tiles.size()
&&PayloadRows.size()==Desired.Tiles.size()
&&TopologyRows.size()==Desired.Tiles.size();
std::vector<FHashedTile> FullDensity,FullMesh,FullPayload,FullTopology;
for(const FHashedTile& Row:DensityRows)if(Row.Key.Level<World.FullResClipLevels)FullDensity.push_back(Row);
for(const FHashedTile& Row:MeshRows)if(Row.Key.Level<World.FullResClipLevels)FullMesh.push_back(Row);
for(const FHashedTile& Row:PayloadRows)if(Row.Key.Level<World.FullResClipLevels)FullPayload.push_back(Row);
for(const FHashedTile& Row:TopologyRows)if(Row.Key.Level<World.FullResClipLevels)FullTopology.push_back(Row);
std::printf("[PipelineFullDetailHash] tiles=%d density=0x%016llx mesh=0x%016llx payload=0x%016llx topology=0x%016llx\n",
FullDetailExpected,static_cast<unsigned long long>(HashTileRows(FullDensity)),
static_cast<unsigned long long>(HashTileRows(FullMesh)),
static_cast<unsigned long long>(HashTileRows(FullPayload)),
static_cast<unsigned long long>(HashTileRows(FullTopology)));
std::printf("[PipelineWholeViewHash] tiles=%zu density=0x%016llx mesh=0x%016llx payload=0x%016llx topology=0x%016llx\n",
Desired.Tiles.size(),static_cast<unsigned long long>(HashTileRows(DensityRows)),
static_cast<unsigned long long>(HashTileRows(MeshRows)),
static_cast<unsigned long long>(HashTileRows(PayloadRows)),
static_cast<unsigned long long>(HashTileRows(TopologyRows)));
for(std::size_t LOD=0;LOD<TileCountByLOD.size();++LOD)
{
const auto FilterLOD=[LOD](const std::vector<FHashedTile>& Rows)
{
std::vector<FHashedTile> Selected;
for(const FHashedTile& Row:Rows)
if(Row.Key.Level==static_cast<int32>(LOD))Selected.push_back(Row);
return Selected;
};
auto LodDensity=FilterLOD(DensityRows),LodMesh=FilterLOD(MeshRows);
auto LodPayload=FilterLOD(PayloadRows),LodTopology=FilterLOD(TopologyRows);
std::printf("[PipelineLODHash] lod=%zu tiles=%llu density=0x%016llx mesh=0x%016llx payload=0x%016llx topology=0x%016llx\n",
LOD,static_cast<unsigned long long>(TileCountByLOD[LOD]),
static_cast<unsigned long long>(HashTileRows(LodDensity)),
static_cast<unsigned long long>(HashTileRows(LodMesh)),
static_cast<unsigned long long>(HashTileRows(LodPayload)),
static_cast<unsigned long long>(HashTileRows(LodTopology)));
}
const std::array<VoxelForgeCore::Vec3d,9> QueryPoints={{
{CameraPositionVoxels.X,CameraPositionVoxels.Y,CameraPositionVoxels.Z},
{CameraPositionVoxels.X+8.0,CameraPositionVoxels.Y,CameraPositionVoxels.Z},
{CameraPositionVoxels.X-8.0,CameraPositionVoxels.Y,CameraPositionVoxels.Z},
{CameraPositionVoxels.X,CameraPositionVoxels.Y+8.0,CameraPositionVoxels.Z},
{CameraPositionVoxels.X,CameraPositionVoxels.Y-8.0,CameraPositionVoxels.Z},
{CameraPositionVoxels.X,CameraPositionVoxels.Y,CameraPositionVoxels.Z+8.0},
{CameraPositionVoxels.X,CameraPositionVoxels.Y,CameraPositionVoxels.Z-8.0},
{CameraPositionVoxels.X+64.0,CameraPositionVoxels.Y-32.0,CameraPositionVoxels.Z+16.0},
{CameraPositionVoxels.X-128.0,CameraPositionVoxels.Y+96.0,CameraPositionVoxels.Z-64.0}}};
uint64 ExactDensityHash=14695981039346656037ull;
uint64 ExactUnclampedHash=14695981039346656037ull;
uint64 ExactDetailHash=14695981039346656037ull;
uint64 ExactCollisionHash=14695981039346656037ull;
for(const auto& Point:QueryPoints)
{
const float Density=Shape.densityAt(Point);
const float Unclamped=Shape.unclampedDensityAt(Point);
const auto Detail=Shape.detailSampleAt(Point);
HashWord(ExactDensityHash,std::bit_cast<uint32>(Density),4);
HashWord(ExactUnclampedHash,std::bit_cast<uint32>(Unclamped),4);
HashWord(ExactDetailHash,std::bit_cast<uint32>(Detail.displacement),4);
HashWord(ExactDetailHash,std::bit_cast<uint32>(Detail.caveInfluence),4);
HashWord(ExactDetailHash,std::bit_cast<uint32>(Detail.floorInfluence),4);
HashWord(ExactCollisionHash,Shape.isSolidAt(Point)?1u:0u,1);
}
std::printf("[PipelineExactQueryHash] probes=%zu density=0x%016llx unclamped=0x%016llx detail=0x%016llx collision=0x%016llx path=direct-Core-queries\n",
QueryPoints.size(),static_cast<unsigned long long>(ExactDensityHash),
static_cast<unsigned long long>(ExactUnclampedHash),
static_cast<unsigned long long>(ExactDetailHash),
static_cast<unsigned long long>(ExactCollisionHash));
}
return NeighborhoodFinished==static_cast<int32>(Neighborhood.size())
&&FullDetailFinished==FullDetailExpected&&Finished==Update.DesiredTiles&&bHashesPassing
&&(!bProfileColumnOverlap||bColumnOverlapCapturePassing)
&&(!bExportRequested||bSceneExportPassing)&&bScreenSpaceGatePassing?0:6;
}
} // namespace StandaloneStreamBenchmark
Reference in new issue View Git Blame Copy Permalink