public void Init()
{
rng = new Random(1);
f = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * iterations, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
f[i] = new float2(0.0f);
}
}
/// <summary>
/// Insert
/// </summary>
public Insertion(Entity obstacle, float4x4 ltw, float2 *vertices, int amount)
{
Type = InsertionType.Insert;
Obstacle = obstacle;
Ltw = ltw;
Vertices = vertices;
Amount = amount;
Amounts = default;
}
/// <summary>
/// Bulk Insert
/// </summary>
public TreeOperation(float4x4 ltw, float2 *verts, int *amounts, int length)
{
Type = TreeOperationType.BulkInsert;
Obstacle = default;
Ltw = ltw;
Vertices = verts;
Amount = length;
Amounts = amounts;
}
/// <summary>
/// Destroy
/// </summary>
public TreeOperation(Entity obstacle)
{
Type = TreeOperationType.Destroy;
Obstacle = obstacle;
Ltw = default;
Vertices = default;
Amount = default;
Amounts = default;
}
public void InsertObstacle(Entity key, float4x4 ltw, float2 *vertices, int amount)
{
Assert.IsTrue(amount > 1);
Assert.IsTrue(key != Entity.Null);
Assert.IsTrue(!Map.ContainsKey(key));
Assert.IsTrue(amount > 2 && math.all(vertices[0] == vertices[amount - 1]), "Obstacle needs to be counter cockwise wound and closed");
Obstacle *first = default;
Obstacle *previous = default;
var prevPos = Math.Mul2D(ltw, vertices[0]);
var prevPos2 = Math.Mul2D(ltw, vertices[amount - 2]);
for (var i = 0; i < amount - 1; ++i)
{
var obstacle = ObstaclePool.GetElementPointer();
obstacle->Point = prevPos;
var pos = Math.Mul2D(ltw, vertices[i + 1]);
var aabb = AABB.FromOpposingCorners(prevPos, pos);
obstacle->Id = Tree.Insert(aabb, (IntPtr)obstacle);
obstacle->Direction = math.normalize(pos - prevPos);
if (amount == 2)
{
obstacle->Convex = true;
}
else
{
obstacle->Convex = LeftOf(prevPos2, prevPos, pos) >= 0;
}
prevPos2 = prevPos;
prevPos = pos;
if (i == 0)
{
first = obstacle;
previous = obstacle;
}
else if (i < amount - 2)
{
obstacle->Previous = previous;
previous->Next = obstacle;
previous = obstacle;
}
else
{
obstacle->Previous = previous;
previous->Next = obstacle;
first->Previous = obstacle;
obstacle->Next = first;
}
}
Map.Add(key, (IntPtr)first);
public void Execute(int i)
{
Particle p = ps[i];
p.v = 0;
uint2 cell_idx = (uint2)p.pos;
float2 cell_diff = (p.pos - cell_idx) - 0.5f;
float2 *w = stackalloc float2[3];
w[0] = 0.5f * math.pow(0.5f - cell_diff, 2);
w[1] = 0.75f - math.pow(cell_diff, 2);
w[2] = 0.5f * math.pow(0.5f + cell_diff, 2);
float2x2 B = 0;
for (uint x = 0; x < 3; ++x)
{
for (uint y = 0; y < 3; ++y)
{
float weight = w[x].x * w[y].y;
uint2 current_cell_idx = math.uint2(cell_idx.x + x - 1, cell_idx.y + y - 1);
float2 cell_dist = (current_cell_idx - p.pos) + 0.5f;
int index_1d = (int)current_cell_idx.x * grid_res + (int)current_cell_idx.y;
float2 weighted_velocity = grid[index_1d].v * weight;
float2x2 term = math.float2x2(weighted_velocity * cell_dist.x, weighted_velocity * cell_dist.y);
B += term;
p.v += weighted_velocity;
}
}
p.C = B * 4;
p.pos += p.v * dt;
p.pos = math.clamp(p.pos, 1, grid_res - 2);
float2 x_n = p.pos + p.v;
float wall_min = 3 - 1;
float wall_max = (float)grid_res - 4 + 1;
p.v.x += (x_n.x < wall_min) ? wall_min - x_n.x : 0;
p.v.x += (x_n.x > wall_max) ? wall_max - x_n.x : 0;
p.v.y += (x_n.y < wall_min) ? wall_min - x_n.y : 0;
p.v.y += (x_n.y > wall_max) ? wall_max - x_n.y : 0;
// if (x_n.x < wall_min) p.v.x += wall_min - x_n.x;
// if (x_n.x > wall_max) p.v.x += wall_max - x_n.x;
// if (x_n.y < wall_min) p.v.y += wall_min - x_n.y;
// if (x_n.y > wall_max) p.v.y += wall_max - x_n.y;
ps[i] = p;
}
public void Execute(int i)
{
Particle p = particles[i];
// reset particle velocity
p.v = 0;
uint2 cell_index = (uint2)p.pos;
float2 cell_diff = (p.pos - cell_index) - 0.5f;
float2 *w = stackalloc float2[3];
w[0] = 0.5f * math.pow(0.5f - cell_diff, 2);
w[1] = 0.75f - math.pow(cell_diff, 2);
w[2] = 0.5f * math.pow(0.5f + cell_diff, 2);
// C = B * (D^-1), D=1/4 * (delta_x)^2 * I when using quadratic interpolation
float2x2 B = 0;
for (uint x = 0; x < 3; ++x)
{
for (uint y = 0; y < 3; ++y)
{
float weight = w[x].x * w[y].y;
uint2 current_index = math.uint2(cell_index.x + x - 1, cell_index.y + y - 1);
int index = (int)current_index.x * grid_res + (int)current_index.y;
float2 dist = (current_index - p.pos) + 0.5f;
float2 weighted_velocity = grids[index].v * weight;
// APIC paper equation 10, constructing inner term for B
float2x2 term = math.float2x2(weighted_velocity * dist.x, weighted_velocity * dist.y);
B += term;
p.v += weighted_velocity;
}
}
p.C = B * 4;
p.pos += p.v * dt;
p.pos = math.clamp(p.pos, 1, grid_res - 2);
//deformation gradient update eqation-181
float2x2 F_new = math.float2x2(
1, 0,
0, 1
);
F_new += dt * p.C;
Fs[i] = math.mul(F_new, Fs[i]);
particles[i] = p;
}
public void Execute(int i)
{
Particle p = particles[i];
float2x2 F = Fs[i];
float2x2 stress = 0;
float J = math.determinant(F);
float volume = p.volume0 * J;
float2x2 F_T = math.transpose(F);
float2x2 inv_F_T = math.inverse(F_T);
float2x2 F_minus_inv_F_T = F - inv_F_T;
float2x2 P = elastic_mu * F_minus_inv_F_T + elastic_lambda * math.log(J) * inv_F_T;
stress = (1.0f / J) * math.mul(P, F_T);
float2x2 eq_16_term_0 = -volume * 4 * stress * dt;
uint2 cell_index = (uint2)p.pos;
float2 cell_diff = (p.pos - cell_index) - 0.5f;
float2 *w = stackalloc float2[3];
w[0] = 0.5f * math.pow(0.5f - cell_diff, 2);
w[1] = 0.75f - math.pow(cell_diff, 2);
w[2] = 0.5f * math.pow(0.5f + cell_diff, 2);
for (uint x = 0; x < 3; ++x)
{
for (uint y = 0; y < 3; ++y)
{
float weight = w[x].x * w[y].y;
uint2 current_index = math.uint2(cell_index.x + x - 1, cell_index.y + y - 1);
float2 dist = (current_index - p.pos) + 0.5f;
float2 Q = math.mul(p.C, dist);
int index_1d = (int)current_index.x * grid_res + (int)current_index.y;
Cell c = grids[index_1d];
float mass_contribute = weight * p.mass;
c.mass += mass_contribute;
c.v += mass_contribute * (p.v + Q);
float2 momentum = math.mul(eq_16_term_0 * weight, dist);
c.v += momentum;
// current cell.v is w_ij * (dt * M^-1 * p.volume * p.stress + p.mass * p.C)
grids[index_1d] = c;
}
}
}
public void Init()
{
rng = (Random *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <Random>() * 10000, UnsafeUtility.AlignOf <Random>(), Allocator.Persistent);
for (int i = 0; i < 10000; ++i)
{
rng[i] = new Unity.Mathematics.Random(1);
}
f = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 10000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 10000; ++i)
{
f[i] = new float2(0.0f);
}
}
public void Init()
{
m1 = (float2x2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2x2>() * 10000, UnsafeUtility.AlignOf <float2x2>(), Allocator.Persistent);
for (int i = 0; i < 10000; ++i)
{
m1[i] = float2x2.identity;
}
m2 = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 10000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 10000; ++i)
{
m2[i] = new float2(1.0f, 0.0f);
}
}
public void Init()
{
sin = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 10000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 10000; ++i)
{
sin[i] = new float2(0.0f);
}
cos = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 10000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 10000; ++i)
{
cos[i] = new float2(1.0f);
}
}
public void Init()
{
v = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 100000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 100000; ++i)
{
v[i] = new float2(1.0f);
}
hash = (uint *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <uint>() * 100000, UnsafeUtility.AlignOf <uint>(), Allocator.Persistent);
for (int i = 0; i < 100000; ++i)
{
hash[i] = 0;
}
}
private static unsafe void TessellateBurst(Allocator allocator, float2 *points, int pointCount, int2 *edges, int edgeCount, float2 *outVertices, int *outIndices, int2 *outEdges, int arrayCount, int3 *result)
{
NativeArray <int2> _edges = new NativeArray <int2>(edgeCount, allocator);
for (int i = 0; i < _edges.Length; ++i)
{
_edges[i] = edges[i];
}
NativeArray <float2> _points = new NativeArray <float2>(pointCount, allocator);
for (int i = 0; i < _points.Length; ++i)
{
_points[i] = points[i];
}
NativeArray <int> _outIndices = new NativeArray <int>(arrayCount, allocator);
NativeArray <int2> _outEdges = new NativeArray <int2>(arrayCount, allocator);
NativeArray <float2> _outVertices = new NativeArray <float2>(arrayCount, allocator);
int outEdgeCount = 0;
int outIndexCount = 0;
int outVertexCount = 0;
ModuleHandle.Tessellate(allocator, _points, _edges, ref _outVertices, ref outVertexCount, ref _outIndices, ref outIndexCount, ref _outEdges, ref outEdgeCount);
for (int i = 0; i < outEdgeCount; ++i)
{
outEdges[i] = _outEdges[i];
}
for (int i = 0; i < outIndexCount; ++i)
{
outIndices[i] = _outIndices[i];
}
for (int i = 0; i < outVertexCount; ++i)
{
outVertices[i] = _outVertices[i];
}
result->x = outVertexCount;
result->y = outIndexCount;
result->z = outEdgeCount;
_outVertices.Dispose();
_outEdges.Dispose();
_outIndices.Dispose();
_points.Dispose();
_edges.Dispose();
}
public void Init()
{
rng = new Random(1);
v = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * iterations, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
v[i] = new float2(1.0f);
}
hash = (uint *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <uint>() * iterations, UnsafeUtility.AlignOf <uint>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
hash[i] = 0;
}
}
public void Init()
{
rng = new Random(1);
m1 = (float2x2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2x2>() * iterations, UnsafeUtility.AlignOf <float2x2>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
m1[i] = float2x2.identity;
}
m2 = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * iterations, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
m2[i] = new float2(1.0f, 0.0f);
}
}
private static unsafe void SubdivideBurst(Allocator allocator, float2 *points, int pointCount, int2 *edges, int edgeCount, float2 *outVertices, int *outIndices, int2 *outEdges, int arrayCount, float areaFactor, float areaThreshold, int refineIterations, int smoothenIterations, int3 *result)
{
NativeArray <int2> _edges = new NativeArray <int2>(edgeCount, allocator);
for (int i = 0; i < _edges.Length; ++i)
{
_edges[i] = edges[i];
}
NativeArray <float2> _points = new NativeArray <float2>(pointCount, allocator);
for (int i = 0; i < _points.Length; ++i)
{
_points[i] = points[i];
}
NativeArray <int> _outIndices = new NativeArray <int>(arrayCount, allocator);
NativeArray <int2> _outEdges = new NativeArray <int2>(arrayCount, allocator);
NativeArray <float2> _outVertices = new NativeArray <float2>(arrayCount, allocator);
int outEdgeCount = 0;
int outIndexCount = 0;
int outVertexCount = 0;
ModuleHandle.Subdivide(allocator, _points, _edges, ref _outVertices, ref outVertexCount, ref _outIndices, ref outIndexCount, ref _outEdges, ref outEdgeCount, areaFactor, areaThreshold, refineIterations, smoothenIterations);
for (int i = 0; i < outEdgeCount; ++i)
{
outEdges[i] = _outEdges[i];
}
for (int i = 0; i < outIndexCount; ++i)
{
outIndices[i] = _outIndices[i];
}
for (int i = 0; i < outVertexCount; ++i)
{
outVertices[i] = _outVertices[i];
}
result->x = outVertexCount;
result->y = outIndexCount;
result->z = outEdgeCount;
_outVertices.Dispose();
_outEdges.Dispose();
_outIndices.Dispose();
_points.Dispose();
_edges.Dispose();
}
public MTerrainBoundingTree(int treeLevel)
{
isCreate = false;
this.treeLevel = treeLevel;
offset = MUnsafeUtility.Malloc <long>(sizeof(long) * treeLevel, Unity.Collections.Allocator.Persistent);
offset[0] = 0;
int j = 1;
for (int i = 1; i < treeLevel; ++i, j *= 2)
{
int ofst = (int)(0.1 + pow(2, i));
offset[i] = offset[i - 1] + j * j;
}
byteSize = (offset[treeLevel - 1] + j * j) * sizeof(float2);
boundingValue = MUnsafeUtility.Malloc <float2>(byteSize, Unity.Collections.Allocator.Persistent);
}
public void Init()
{
v1 = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 100000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 100000; ++i)
{
v1[i] = new float2(1.0f);
}
v2 = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 100000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 100000; ++i)
{
v2[i] = new float2(2.0f);
}
result = (float2 *)UnsafeUtility.Malloc(UnsafeUtility.SizeOf <float2>() * 100000, UnsafeUtility.AlignOf <float2>(), Allocator.Persistent);
for (int i = 0; i < 100000; ++i)
{
result[i] = new float2(1.0f);
}
}
public static bool IntersectTriangle(float3 orig, float3 dir, float3 v0, float3 v1, float3 v2
, float *t, float2 *uv)
{
float3 e1 = v1 - v0;
float3 e2 = v2 - v0;
float3 p = cross(dir, e2);
float det = dot(e1, p);
float3 T;
if (det > 0)
{
T = orig - v0;
}
else
{
T = v0 - orig;
det = -det;
}
if (det < 0.0001)
{
return(false);
}
uv->x = dot(T, p);
if (uv->x < 0.0f || uv->x > det)
{
return(false);
}
float3 Q = cross(T, e1);
uv->y = dot(dir, Q);
if (uv->y < 0.0f || (uv->x + uv->y) > det)
{
return(false);
}
*t = dot(e2, Q);
float fInvDet = 1 / det;
*t *= fInvDet;
*uv *= fInvDet;
return(true);
}
public static Mesh GeneratePlaneMeshFromCoord(NativeList <float4> coords, float4x4 localToModelMatrix, float2 startUV, float2 endUV)
{
List <Vector3> allVertices = new List <Vector3>(coords.Length * 4);
List <Vector2> uvs = new List <Vector2>(allVertices.Capacity);
List <Vector3> normals = new List <Vector3>(allVertices.Capacity);
List <Vector4> tans = new List <Vector4>(allVertices.Capacity);
List <int> triangles = new List <int>(allVertices.Capacity * 3 / 2 + 1);
void AddPoint(float2 leftUpCorner, float2 rightDownCorner)
{
int * triangleCount = stackalloc int[] { 0, 1, 2, 1, 3, 2 };
float2 *verts = stackalloc float2[] { leftUpCorner, float2(rightDownCorner.x, leftUpCorner.y), float2(leftUpCorner.x, rightDownCorner.y), rightDownCorner };
int len = allVertices.Count;
for (int i = 0; i < 6; ++i)
{
triangles.Add(triangleCount[i] + len);
}
for (int i = 0; i < 4; ++i)
{
allVertices.Add(mul(localToModelMatrix, float4(verts[i], 0, 1)).xyz);
uvs.Add(lerp(startUV, endUV, verts[i]));
normals.Add(mul(localToModelMatrix, float4(0, 0, 1, 0)).xyz);
tans.Add(float4(mul(localToModelMatrix, float4(1, 0, 0, 0)).xyz, 1));
}
}
Mesh m = new Mesh();
foreach (var i in coords)
{
AddPoint(i.xy, i.zw);
}
m.SetVertices(allVertices);
m.SetUVs(0, uvs);
m.SetTangents(tans);
m.SetNormals(normals);
m.SetTriangles(triangles, 0);
return(m);
}
public void Init()
{
rng = new Random(1);
v1 = (float2*)UnsafeUtility.Malloc(UnsafeUtility.SizeOf<float2>() * iterations, UnsafeUtility.AlignOf<float2>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
v1[i] = new float2(1.0f);
}
v2 = (float2*)UnsafeUtility.Malloc(UnsafeUtility.SizeOf<float2>() * iterations, UnsafeUtility.AlignOf<float2>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
v2[i] = new float2(2.0f);
}
result = (float2*)UnsafeUtility.Malloc(UnsafeUtility.SizeOf<float2>() * iterations, UnsafeUtility.AlignOf<float2>(), Allocator.Persistent);
for (int i = 0; i < iterations; ++i)
{
result[i] = new float2(1.0f);
}
}
public unsafe void Draw()
{
if (VertexCount < 3)
{
return;
}
BeginLineStrip(Color);
fixed(byte *array = Vertices)
{
float2 *vertexArray = (float2 *)array;
for (var i = 0; i < VertexCount; ++i)
{
var vertex = PhysicsMath.mul(Transform, vertexArray[i]);
DrawLineVertex(vertex);
}
DrawLineVertex(PhysicsMath.mul(Transform, vertexArray[0]));
}
EndDraw();
}
unsafe JobHandle?GetUvsJob(
void *input,
int count,
GLTFComponentType inputType,
int inputByteStride,
float2 *output,
int outputByteStride,
bool normalized = false
)
{
Profiler.BeginSample("PrepareUVs");
JobHandle?jobHandle = null;
switch (inputType)
{
case GLTFComponentType.Float:
{
var jobUv = new Jobs.ConvertUVsFloatToFloatInterleavedJob {
inputByteStride = (inputByteStride > 0) ? inputByteStride : 8,
input = (byte *)input,
outputByteStride = outputByteStride,
result = output
};
#if UNITY_JOBS
jobHandle = jobUv.ScheduleBatch(count, GltfImport.DefaultBatchCount);
#else
jobHandle = jobUv.Schedule(count, GltfImport.DefaultBatchCount);
#endif
}
break;
case GLTFComponentType.UnsignedByte:
if (normalized)
{
var jobUv = new Jobs.ConvertUVsUInt8ToFloatInterleavedNormalizedJob {
inputByteStride = (inputByteStride > 0) ? inputByteStride : 2,
input = (byte *)input,
outputByteStride = outputByteStride,
result = output
};
jobHandle = jobUv.Schedule(count, GltfImport.DefaultBatchCount);
}
else
{
var jobUv = new Jobs.ConvertUVsUInt8ToFloatInterleavedJob {
inputByteStride = (inputByteStride > 0) ? inputByteStride : 2,
input = (byte *)input,
outputByteStride = outputByteStride,
result = output
};
#if UNITY_JOBS
jobHandle = jobUv.ScheduleBatch(count, GltfImport.DefaultBatchCount);
#else
jobHandle = jobUv.Schedule(count, GltfImport.DefaultBatchCount);
#endif
}
break;
case GLTFComponentType.UnsignedShort:
if (normalized)
{
var jobUv = new Jobs.ConvertUVsUInt16ToFloatInterleavedNormalizedJob {
inputByteStride = (inputByteStride > 0) ? inputByteStride : 4,
input = (byte *)input,
outputByteStride = outputByteStride,
result = output
};
jobHandle = jobUv.Schedule(count, GltfImport.DefaultBatchCount);
}
else
{
var jobUv = new Jobs.ConvertUVsUInt16ToFloatInterleavedJob {
inputByteStride = (inputByteStride > 0) ? inputByteStride : 4,
input = (byte *)input,
outputByteStride = outputByteStride,
result = output
};
#if UNITY_JOBS
jobHandle = jobUv.ScheduleBatch(count, GltfImport.DefaultBatchCount);
#else
jobHandle = jobUv.Schedule(count, GltfImport.DefaultBatchCount);
#endif
}
break;
case GLTFComponentType.Short:
if (normalized)
{
var job = new Jobs.ConvertUVsInt16ToFloatInterleavedNormalizedJob {
inputByteStride = inputByteStride > 0 ? inputByteStride : 4,
input = (System.Int16 *)input,
outputByteStride = outputByteStride,
result = output
};
#if UNITY_JOBS
jobHandle = job.ScheduleBatch(count, GltfImport.DefaultBatchCount);
#else
jobHandle = job.Schedule(count, GltfImport.DefaultBatchCount);
#endif
}
else
{
var job = new Jobs.ConvertUVsInt16ToFloatInterleavedJob {
inputByteStride = inputByteStride > 0 ? inputByteStride : 4,
input = (System.Int16 *)input,
outputByteStride = outputByteStride,
result = output
};
#if UNITY_JOBS
jobHandle = job.ScheduleBatch(count, GltfImport.DefaultBatchCount);
#else
jobHandle = job.Schedule(count, GltfImport.DefaultBatchCount);
#endif
}
break;
case GLTFComponentType.Byte:
var byteStride = inputByteStride > 0 ? inputByteStride : 2;
if (normalized)
{
var jobInt8 = new Jobs.ConvertUVsInt8ToFloatInterleavedNormalizedJob {
inputByteStride = inputByteStride > 0 ? inputByteStride : 2,
input = (sbyte *)input,
outputByteStride = outputByteStride,
result = output
};
#if UNITY_JOBS
jobHandle = jobInt8.ScheduleBatch(count, GltfImport.DefaultBatchCount);
#else
jobHandle = jobInt8.Schedule(count, GltfImport.DefaultBatchCount);
#endif
}
else
{
var jobInt8 = new Jobs.ConvertUVsInt8ToFloatInterleavedJob {
inputByteStride = inputByteStride > 0 ? inputByteStride : 2,
input = (sbyte *)input,
outputByteStride = outputByteStride,
result = output
};
#if UNITY_JOBS
jobHandle = jobInt8.ScheduleBatch(count, GltfImport.DefaultBatchCount);
#else
jobHandle = jobInt8.Schedule(count, GltfImport.DefaultBatchCount);
#endif
}
break;
default:
logger?.Error(LogCode.TypeUnsupported, "UV", inputType.ToString());
break;
}
Profiler.EndSample();
return(jobHandle);
}
// This differs from CullIntermediateLights in 3 ways:
// - tile-frustums/light intersection use different algorithm
// - depth range of the light shape intersecting the tile-frustums is output in the tile list header section
// - light indices written out are indexing visible_lights, rather than the array of PrePunctualLights.
unsafe public void CullFinalLights(ref NativeArray <PrePunctualLight> punctualLights,
ref NativeArray <ushort> lightIndices, int lightStartIndex, int lightCount,
int istart, int iend, int jstart, int jend)
{
// Interestingly, 2-3% faster when using unsafe arrays.
PrePunctualLight *_punctualLights = (PrePunctualLight *)NativeArrayUnsafeUtility.GetUnsafeBufferPointerWithoutChecks(punctualLights);
ushort * _lightIndices = (ushort *)NativeArrayUnsafeUtility.GetUnsafeBufferPointerWithoutChecks(lightIndices);
uint * _tileHeaders = (uint *)NativeArrayUnsafeUtility.GetUnsafeBufferPointerWithoutChecks(m_TileHeaders);
if (lightCount == 0)
{
for (int j = jstart; j < jend; ++j)
{
for (int i = istart; i < iend; ++i)
{
int headerOffset = GetTileHeaderOffset(i, j);
_tileHeaders[headerOffset + 0] = 0;
_tileHeaders[headerOffset + 1] = 0;
_tileHeaders[headerOffset + 2] = 0;
_tileHeaders[headerOffset + 3] = 0;
}
}
return;
}
// Store culled lights in temporary buffer. Additionally store depth range of each light for a given tile too.
// the depth range is a 32bit mask, but packed into a 16bits value since the range of the light is continuous
// (only need to store first bit enabled, and count of enabled bits).
ushort *tiles = stackalloc ushort[lightCount * 2];
float2 *depthRanges = stackalloc float2[lightCount];
int maxLightPerTile = 0; // for stats
int lightEndIndex = lightStartIndex + lightCount;
float2 tileSize = new float2((m_FrustumPlanes.right - m_FrustumPlanes.left) / m_TileXCount, (m_FrustumPlanes.top - m_FrustumPlanes.bottom) / m_TileYCount);
float2 tileExtents = tileSize * 0.5f;
float2 tileExtentsInv = new float2(1.0f / tileExtents.x, 1.0f / tileExtents.y);
for (int j = jstart; j < jend; ++j)
{
float tileYCentre = m_FrustumPlanes.top - (tileExtents.y + j * tileSize.y);
for (int i = istart; i < iend; ++i)
{
float tileXCentre = m_FrustumPlanes.left + tileExtents.x + i * tileSize.x;
PreTile preTile = m_PreTiles[i + j * m_TileXCount];
int culledLightCount = 0;
// For the current tile's light list, min&max depth range (absolute values).
float listMinDepth = float.MaxValue;
float listMaxDepth = -float.MaxValue;
// Duplicate the inner loop twice. Testing for the ortographic case inside the inner loop would cost an extra 8% otherwise.
// Missing C++ template argument here!
if (!m_IsOrthographic)
{
for (int vi = lightStartIndex; vi < lightEndIndex; ++vi)
{
ushort lightIndex = _lightIndices[vi];
PrePunctualLight ppl = _punctualLights[lightIndex];
// Offset tileCentre toward the light to calculate a more conservative minMax depth bound,
// but it must remains inside the tile and must not pass further than the light centre.
float2 tileCentre = new float2(tileXCentre, tileYCentre);
float2 dir = ppl.screenPos - tileCentre;
float2 d = abs(dir * tileExtentsInv);
float sInv = 1.0f / max3(d.x, d.y, 1.0f);
float3 tileOffCentre = new float3(tileCentre.x + dir.x * sInv, tileCentre.y + dir.y * sInv, -m_FrustumPlanes.zNear);
float3 tileOrigin = new float3(0.0f);
float t0, t1;
// This is more expensive than Clip() but allow to compute min&max depth range for the part of the light inside the tile.
if (!IntersectionLineSphere(ppl.posVS, ppl.radius, tileOrigin, tileOffCentre, out t0, out t1))
{
continue;
}
listMinDepth = listMinDepth < t0 ? listMinDepth : t0;
listMaxDepth = listMaxDepth > t1 ? listMaxDepth : t1;
depthRanges[culledLightCount] = new float2(t0, t1);
// Because this always output to the finest tiles, contrary to CullLights(),
// the result are indices into visibleLights, instead of indices into punctualLights.
tiles[culledLightCount] = ppl.visLightIndex;
++culledLightCount;
}
}
else
{
for (int vi = lightStartIndex; vi < lightEndIndex; ++vi)
{
ushort lightIndex = _lightIndices[vi];
PrePunctualLight ppl = _punctualLights[lightIndex];
// Offset tileCentre toward the light to calculate a more conservative minMax depth bound,
// but it must remains inside the tile and must not pass further than the light centre.
float2 tileCentre = new float2(tileXCentre, tileYCentre);
float2 dir = ppl.screenPos - tileCentre;
float2 d = abs(dir * tileExtentsInv);
float sInv = 1.0f / max3(d.x, d.y, 1.0f);
float3 tileOffCentre = new float3(0, 0, -m_FrustumPlanes.zNear);
float3 tileOrigin = new float3(tileCentre.x + dir.x * sInv, tileCentre.y + dir.y * sInv, 0.0f);
float t0, t1;
// This is more expensive than Clip() but allow to compute min&max depth range for the part of the light inside the tile.
if (!IntersectionLineSphere(ppl.posVS, ppl.radius, tileOrigin, tileOffCentre, out t0, out t1))
{
continue;
}
listMinDepth = listMinDepth < t0 ? listMinDepth : t0;
listMaxDepth = listMaxDepth > t1 ? listMaxDepth : t1;
depthRanges[culledLightCount] = new float2(t0, t1);
// Because this always output to the finest tiles, contrary to CullLights(),
// the result are indices into visibleLights, instead of indices into punctualLights.
tiles[culledLightCount] = ppl.visLightIndex;
++culledLightCount;
}
}
// Post-multiply by zNear to get actual world unit absolute depth values, then clamp to valid depth range.
listMinDepth = max2(listMinDepth * m_FrustumPlanes.zNear, m_FrustumPlanes.zNear);
listMaxDepth = min2(listMaxDepth * m_FrustumPlanes.zNear, m_FrustumPlanes.zFar);
// Calculate bitmask for 2.5D culling.
uint bitMask = 0;
float depthRangeInv = 1.0f / (listMaxDepth - listMinDepth);
for (int culledLightIndex = 0; culledLightIndex < culledLightCount; ++culledLightIndex)
{
float lightMinDepth = max2(depthRanges[culledLightIndex].x * m_FrustumPlanes.zNear, m_FrustumPlanes.zNear);
float lightMaxDepth = min2(depthRanges[culledLightIndex].y * m_FrustumPlanes.zNear, m_FrustumPlanes.zFar);
int firstBit = (int)((lightMinDepth - listMinDepth) * 32.0f * depthRangeInv);
int lastBit = (int)((lightMaxDepth - listMinDepth) * 32.0f * depthRangeInv);
int bitCount = min(lastBit - firstBit + 1, 32 - firstBit);
bitMask |= (uint)((0xFFFFFFFF >> (32 - bitCount)) << firstBit);
tiles[culledLightCount + culledLightIndex] = (ushort)((uint)firstBit | (uint)(bitCount << 8));
}
// As listMinDepth and listMaxDepth are used to calculate the geometry 2.5D bitmask,
// we can optimize the shader execution (TileDepthInfo.shader) by refactoring the calculation.
// int bitIndex = 32.0h * (geoDepth - listMinDepth) / (listMaxDepth - listMinDepth);
// Equivalent to:
// a = 32.0 / (listMaxDepth - listMinDepth)
// b = -listMinDepth * 32.0 / (listMaxDepth - listMinDepth)
// int bitIndex = geoDepth * a + b;
float a = 32.0f * depthRangeInv;
float b = -listMinDepth * a;
int tileDataSize = culledLightCount * 2;
int tileOffset = culledLightCount > 0 ? AddTileData(tiles, ref tileDataSize) : 0;
int headerOffset = GetTileHeaderOffset(i, j);
_tileHeaders[headerOffset + 0] = (uint)tileOffset;
_tileHeaders[headerOffset + 1] = (uint)(tileDataSize == 0 ? 0 : culledLightCount);
_tileHeaders[headerOffset + 2] = _f32tof16(a) | (_f32tof16(b) << 16);
_tileHeaders[headerOffset + 3] = bitMask;
maxLightPerTile = max(maxLightPerTile, culledLightCount);
}
}
m_Counters[0] = max(m_Counters[0], maxLightPerTile); // TODO make it atomic
}
protected static extern unsafe int DetailedOverlapInformationNative(IntPtr emr, Entity e, HitBoxOverlap overlap, float2 *result);
public TerrainDrawStreaming(int maximumLength, int meshSize, ComputeShader transformShader)
{
current = this;
if (meshSize % 2 != 0)
{
Debug.LogError("Terrain panel's size should be even number!");
meshSize++;
}
this.meshSize = meshSize;
//Initialize Mesh and triangles
int vertexCount = meshSize + 1;
vertSize = vertexCount;
heightMapSize = vertSize * vertSize;
NativeArray <float2> terrainVertexArray = new NativeArray <float2>(vertexCount * vertexCount, Allocator.Temp, NativeArrayOptions.UninitializedMemory);
float2 *arrPtr = terrainVertexArray.Ptr();
for (int x = 0; x < vertexCount; ++x)
{
for (int y = 0; y < vertexCount; ++y)
{
arrPtr[y * vertexCount + x] = new float2(x, y) / meshSize - new float2(0.5f, 0.5f);
}
}
verticesBuffer = new ComputeBuffer(terrainVertexArray.Length, sizeof(float2));
verticesBuffer.SetData(terrainVertexArray);
heightMapBuffer = new ComputeBuffer(maximumLength * (vertexCount * vertexCount), sizeof(float));
NativeArray <int> triangles = new NativeArray <int>(6 * meshSize * meshSize, Allocator.Temp, NativeArrayOptions.UninitializedMemory);
int *trianglePtr = triangles.Ptr();
for (int x = 0, count = 0; x < meshSize; ++x)
{
for (int y = 0; y < meshSize; ++y)
{
int4 indices = new int4(vertexCount * y + x, vertexCount * (y + 1) + x, vertexCount * y + (x + 1), vertexCount * (y + 1) + (x + 1));
trianglePtr[count] = indices.x;
trianglePtr[count + 1] = indices.y;
trianglePtr[count + 2] = indices.z;
trianglePtr[count + 3] = indices.y;
trianglePtr[count + 4] = indices.w;
trianglePtr[count + 5] = indices.z;
count += 6;
}
}
triangleBuffer = new ComputeBuffer(triangles.Length, sizeof(int));
triangleBuffer.SetData(triangles);
triangles.Dispose();
terrainVertexArray.Dispose();
removebuffer = new ComputeBuffer(100, sizeof(int2));
//Initialize indirect
clusterBuffer = new ComputeBuffer(maximumLength, sizeof(TerrainPanel));
referenceBuffer = new NativeList <ulong>(maximumLength, Allocator.Persistent);
this.transformShader = transformShader;
resultBuffer = new ComputeBuffer(maximumLength, sizeof(int));
instanceCountBuffer = new ComputeBuffer(5, sizeof(int), ComputeBufferType.IndirectArguments);
NativeArray <int> indirect = new NativeArray <int>(5, Allocator.Temp, NativeArrayOptions.ClearMemory);
indirect[0] = triangleBuffer.count;
instanceCountBuffer.SetData(indirect);
indirect.Dispose();
notUsedHeightmapIndices = new NativeList <int>(maximumLength, Allocator.Persistent);
for (int i = 0; i < maximumLength; ++i)
{
notUsedHeightmapIndices.Add(i);
}
}
private async void Movie_CTF2DChanged(object sender, EventArgs e)
{
try
{
AdjustGridVisibility();
ImageSimulated2D.Visibility = Visibility.Hidden;
if (Movie == null)
{
return;
}
if (Movie.OptionsCTF == null || Movie.CTF == null)
{
return;
}
this.Width = Movie.OptionsCTF.Window;
this.Height = Movie.OptionsCTF.Window;
ProgressCTF2D.Visibility = Visibility.Visible;
int Width = Movie.OptionsCTF.Window / 2;
CTF MovieCTF = Movie.CTF;
//await Task.Delay(1000);
await Task.Run(() =>
{
ImageSource Simulated2D;
unsafe
{
float2[] SimCoords = new float2[Width *Width];
fixed(float2 *SimCoordsPtr = SimCoords)
{
float2 *SimCoordsP = SimCoordsPtr;
for (int y = 0; y < Width; y++)
{
int ycoord = Width - 1 - y;
int ycoord2 = ycoord *ycoord;
for (int x = 0; x < Width; x++)
{
int xcoord = x - Width;
*SimCoordsP++ = new float2((float)Math.Sqrt(xcoord *xcoord + ycoord2) / (Width * 2), (float)Math.Atan2(ycoord, xcoord));
}
}
}
float[] Sim2D = MovieCTF.Get2D(SimCoords, true, true, true);
byte[] Sim2DBytes = new byte[Sim2D.Length];
fixed(byte *Sim2DBytesPtr = Sim2DBytes)
fixed(float *Sim2DPtr = Sim2D)
{
byte *Sim2DBytesP = Sim2DBytesPtr;
float *Sim2DP = Sim2DPtr;
for (int i = 0; i < Width *Width; i++)
{
*Sim2DBytesP++ = (byte)(*Sim2DP++ *128f + 127f);
}
}
Simulated2D = BitmapSource.Create(Width, Width, 96, 96, PixelFormats.Indexed8, BitmapPalettes.Gray256, Sim2DBytes, Width);
Simulated2D.Freeze();
}
Dispatcher.Invoke(() =>
{
ImageSimulated2D.Source = Simulated2D;
ImageSimulated2D.Visibility = Visibility.Visible;
});
});
ProgressCTF2D.Visibility = Visibility.Hidden;
}
catch
{
}
}
/// <summary>
/// Creates a new CudaRegisteredHostMemory_float2 from an existing IntPtr. IntPtr must be page size aligned (4KBytes)!
/// </summary>
/// <param name="hostPointer">must be page size aligned (4KBytes)</param>
/// <param name="size">In elements</param>
public CudaRegisteredHostMemory_float2(IntPtr hostPointer, SizeT size)
{
_intPtr = hostPointer;
_size = size;
_typeSize = (SizeT)Marshal.SizeOf(typeof(float2));
_ptr = (float2*)_intPtr;
}
public static unsafe void PopulateSpriteVertices(ref WorldSpaceMask rectMask, ref DynamicBuffer <ControlVertexData> vertices,
ref DynamicBuffer <ControlVertexIndex> triangles, ref WorldSpaceRect rectData,
ref SpriteVertexData spriteVertexData, float4 color)
{
float pixelsPerUnit = spriteVertexData.PixelsPerUnit / 100.0f;
Rect rect = new Rect(rectData.Min, rectData.Max - rectData.Min);
float4 adjustedBorders = GetAdjustedBorders(spriteVertexData.Border / pixelsPerUnit, rect);
var padding = spriteVertexData.Padding / pixelsPerUnit;
float2 *vertScratch = stackalloc float2[4];
float2 *uvScratch = stackalloc float2[4];
vertScratch[0] = new float2(padding.x, padding.y);
vertScratch[3] = new float2(rect.width - padding.z, rect.height - padding.w);
vertScratch[1].x = adjustedBorders.x;
vertScratch[1].y = adjustedBorders.y;
vertScratch[2].x = rect.width - adjustedBorders.z;
vertScratch[2].y = rect.height - adjustedBorders.w;
for (int i = 0; i < 4; ++i)
{
vertScratch[i].x += rect.x;
vertScratch[i].y += rect.y;
}
uvScratch[0] = new Vector2(spriteVertexData.Outer.x, spriteVertexData.Outer.y);
uvScratch[1] = new Vector2(spriteVertexData.Inner.x, spriteVertexData.Inner.y);
uvScratch[2] = new Vector2(spriteVertexData.Inner.z, spriteVertexData.Inner.w);
uvScratch[3] = new Vector2(spriteVertexData.Outer.z, spriteVertexData.Outer.w);
// rect mask support
var cut = GetRectangleMaskCut(vertScratch[0], vertScratch[3], ref rectMask);
vertScratch[0] = vertScratch[0] + cut.Min - clamp(cut.Max - float2(rect.size), 0.0f, float.PositiveInfinity);
vertScratch[1] = vertScratch[1] + math.clamp(cut.Min - adjustedBorders.xy, 0.0f, float.PositiveInfinity) - clamp(cut.Max - (float2(rect.size) - adjustedBorders.xy), 0.0f, float.PositiveInfinity);
vertScratch[2] = vertScratch[2] + math.clamp(cut.Min - (float2(rect.size) - adjustedBorders.zw), 0.0f, float.PositiveInfinity) - math.clamp(cut.Max - adjustedBorders.wz, 0.0f, float.PositiveInfinity);
vertScratch[3] = vertScratch[3] + clamp(cut.Min - float2(rect.size), 0.0f, float.PositiveInfinity) - cut.Max;
int vertexOffset = vertices.Length;
for (int x = 0; x < 4; ++x)
{
for (int y = 0; y < 4; ++y)
{
vertices.Add(new ControlVertexData()
{
Position = new float3(vertScratch[x].x, vertScratch[y].y, 0.0f),
TexCoord0 = new float2(uvScratch[x].x, uvScratch[y].y),
Color = color
});
}
}
for (int x = 0; x < 3; ++x)
{
int x2 = x + 1;
for (int y = 0; y < 3; ++y)
{
int y2 = y + 1;
// Ignore empty
if (vertScratch[x2].x - vertScratch[x].x <= 0.0f)
{
continue;
}
if (vertScratch[y2].y - vertScratch[y].y <= 0.0f)
{
continue;
}
triangles.Add(vertexOffset + x * 4 + y);
triangles.Add(vertexOffset + x * 4 + y2);
triangles.Add(vertexOffset + x2 * 4 + y);
triangles.Add(vertexOffset + x2 * 4 + y2);
triangles.Add(vertexOffset + x2 * 4 + y);
triangles.Add(vertexOffset + x * 4 + y2);
}
}
// for (int x = 0; x < 3; ++x)
// {
// int x2 = x + 1;
//
// for (int y = 0; y < 3; ++y)
// {
// int y2 = y + 1;
//
//
// AddQuad(ref rectMask, ref vertices, ref triangles,
// new float2(vertScratch[x].x, vertScratch[y].y),
// new float2(vertScratch[x2].x, vertScratch[y2].y),
// color,
// new float2(uvScratch[x].x, uvScratch[y].y),
// new float2(uvScratch[x2].x, uvScratch[y2].y));
// }
// }
}
public void Execute(int i)
{
float2 * w = stackalloc float2[3];
Particle p = ps[i];
uint2 cell_idx = (uint2)p.pos;
float2 cell_diff = (p.pos - cell_idx) - 0.5f;
w[0] = 0.5f * math.pow(0.5f - cell_diff, 2);
w[1] = 0.75f - math.pow(cell_diff, 2);
w[2] = 0.5f * math.pow(0.5f + cell_diff, 2);
float density = 0.0f;
for (uint x = 0; x < 3; ++x)
{
for (uint y = 0; y < 3; ++y)
{
float weight = w[x].x * w[y].y;
int index_1d = (int)(cell_idx.x + x - 1) * grid_res + (int)(cell_idx.y + y - 1);
density += grid[index_1d].mass * weight; // m_0 / h^3 in this case h = 1
}
}
float volume = p.mass / density;
float pressure = math.max(-0.1f, eos_stiffness * (math.pow(density / rest_density, eos_power) - 1));
float2x2 stress = math.float2x2(
-pressure, 0,
0, -pressure
);
float2x2 velocity_gradient = p.C;
float2x2 velocity_gradient_T = math.transpose(velocity_gradient);
float2x2 strain = velocity_gradient + velocity_gradient_T;
//float trace = strain.c1.x + strain.c0.y;
//strain.c0.y = strain.c1.x = trace;
float2x2 viscosity_term = dynamic_viscosity * strain;
stress += viscosity_term;
float2x2 eq_16_term_0 = -volume * 4 * stress * dt;
for (uint x = 0; x < 3; ++x)
{
for (uint y = 0; y < 3; ++y)
{
float weight = w[x].x * w[y].y;
uint2 current_cell_idx = math.uint2(cell_idx.x + x - 1, cell_idx.y + y - 1);
float2 cell_dist = (current_cell_idx - p.pos) + 0.5f;
int index_1d = (int)current_cell_idx.x * grid_res + (int)current_cell_idx.y;
Cell c = grid[index_1d];
float2 momentum = math.mul(eq_16_term_0 * weight, cell_dist);
c.v += momentum;
grid[index_1d] = c;
}
}
}
