OpenSubdiv/opensubdiv/osd/tbbKernel.cpp

717 lines
27 KiB
C++
Raw Normal View History

2013-06-05 17:55:24 +00:00
//
// Copyright 2013 Pixar
2013-06-05 17:55:24 +00:00
//
// Licensed under the Apache License, Version 2.0 (the "Apache License")
// with the following modification; you may not use this file except in
// compliance with the Apache License and the following modification to it:
// Section 6. Trademarks. is deleted and replaced with:
2013-06-05 17:55:24 +00:00
//
// 6. Trademarks. This License does not grant permission to use the trade
// names, trademarks, service marks, or product names of the Licensor
// and its affiliates, except as required to comply with Section 4(c) of
// the License and to reproduce the content of the NOTICE file.
2013-06-05 17:55:24 +00:00
//
// You may obtain a copy of the Apache License at
2013-06-05 17:55:24 +00:00
//
// http://www.apache.org/licenses/LICENSE-2.0
2013-08-14 16:46:53 +00:00
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the Apache License with the above modification is
// distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the Apache License for the specific
// language governing permissions and limitations under the Apache License.
2013-06-05 17:55:24 +00:00
//
#include "../osd/cpuKernel.h"
#include "../osd/tbbKernel.h"
#include "../osd/vertexDescriptor.h"
#include <math.h>
#include <tbb/parallel_for.h>
namespace OpenSubdiv {
namespace OPENSUBDIV_VERSION {
#define grain_size 200
static inline void
clear(float *origin, int index, OsdVertexBufferDescriptor const &desc) {
if (origin) {
float *dst = origin + index * desc.stride;
memset(dst, 0, desc.length * sizeof(float));
}
}
static inline void
addWithWeight(float *origin, int dstIndex, int srcIndex,
float weight, OsdVertexBufferDescriptor const &desc) {
if (origin) {
const float *src = origin + srcIndex * desc.stride;
float *dst = origin + dstIndex * desc.stride;
for (int k = 0; k < desc.length; ++k) {
dst[k] += src[k] * weight;
}
}
}
2013-06-05 17:55:24 +00:00
class TBBFaceKernel {
float *vertex;
float *varying;
OsdVertexBufferDescriptor vertexDesc;
OsdVertexBufferDescriptor varyingDesc;
int const *F_IT;
int const *F_ITa;
2013-06-05 17:55:24 +00:00
int vertexOffset;
int tableOffset;
public:
void operator() (tbb::blocked_range<int> const &r) const {
if(vertexDesc.length == 4 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeFaceKernel<4>
(vertex, F_IT, F_ITa, vertexOffset, tableOffset, r.begin(), r.end());
} else if(vertexDesc.length == 8 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeFaceKernel<8>
(vertex, F_IT, F_ITa, vertexOffset, tableOffset, r.begin(), r.end());
}
else {
for (int i = r.begin() + tableOffset; i < r.end() + tableOffset; i++) {
int h = F_ITa[2*i];
int n = F_ITa[2*i+1];
float weight = 1.0f/n;
// XXX: should use local vertex struct variable instead of
// accumulating directly into global memory.
int dstIndex = i + vertexOffset - tableOffset;
clear(vertex, dstIndex, vertexDesc);
clear(varying, dstIndex, varyingDesc);
2013-06-05 17:55:24 +00:00
for (int j = 0; j < n; ++j) {
int index = F_IT[h+j];
addWithWeight(vertex, dstIndex, index, weight, vertexDesc);
addWithWeight(varying, dstIndex, index, weight, varyingDesc);
2013-06-05 17:55:24 +00:00
}
}
}
}
2013-06-05 17:55:24 +00:00
TBBFaceKernel(TBBFaceKernel const &other)
2013-06-05 17:55:24 +00:00
{
this->vertex = other.vertex;
this->varying= other.varying;
this->vertexDesc = other.vertexDesc;
this->varyingDesc = other.varyingDesc;
2013-06-05 17:55:24 +00:00
this->F_IT = other.F_IT;
this->F_ITa = other.F_ITa;
this->vertexOffset = other.vertexOffset;
this->tableOffset = other.tableOffset;
}
TBBFaceKernel(float *vertex_in,
2013-06-05 17:55:24 +00:00
float *varying_in,
OsdVertexBufferDescriptor const &vertexDesc_in,
OsdVertexBufferDescriptor const &varyingDesc_in,
int const *F_IT_in,
int const *F_ITa_in,
2013-06-05 17:55:24 +00:00
int vertexOffset_in,
int tableOffset_in) :
vertex (vertex_in),
varying(varying_in),
vertexDesc(vertexDesc_in),
varyingDesc(varyingDesc_in),
2013-06-05 17:55:24 +00:00
F_IT (F_IT_in),
F_ITa (F_ITa_in),
vertexOffset(vertexOffset_in),
tableOffset(tableOffset_in)
{};
2013-06-05 17:55:24 +00:00
};
void OsdTbbComputeFace(
float * vertex, float * varying,
OsdVertexBufferDescriptor const &vertexDesc,
OsdVertexBufferDescriptor const &varyingDesc,
int const *F_IT, int const *F_ITa, int vertexOffset, int tableOffset,
2013-06-05 17:55:24 +00:00
int start, int end) {
TBBFaceKernel kernel(vertex, varying, vertexDesc, varyingDesc, F_IT, F_ITa,
2013-06-05 17:55:24 +00:00
vertexOffset, tableOffset);
tbb::blocked_range<int> range(start, end, grain_size);
2013-06-05 17:55:24 +00:00
tbb::parallel_for(range, kernel);
}
class TBBEdgeKernel {
float *vertex;
float *varying;
OsdVertexBufferDescriptor vertexDesc;
OsdVertexBufferDescriptor varyingDesc;
int const *E_IT;
float const *E_W;
2013-06-05 17:55:24 +00:00
int vertexOffset;
int tableOffset;
public:
void operator() (tbb::blocked_range<int> const &r) const {
if(vertexDesc.length == 4 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeEdgeKernel<4>(vertex, E_IT, E_W, vertexOffset, tableOffset,
r.begin(), r.end());
2013-06-05 17:55:24 +00:00
}
else if(vertexDesc.length == 8 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeEdgeKernel<8>(vertex, E_IT, E_W, vertexOffset, tableOffset,
r.begin(), r.end());
2013-06-05 17:55:24 +00:00
}
else {
for (int i = r.begin() + tableOffset; i < r.end() + tableOffset; i++) {
int eidx0 = E_IT[4*i+0];
int eidx1 = E_IT[4*i+1];
int eidx2 = E_IT[4*i+2];
int eidx3 = E_IT[4*i+3];
float vertWeight = E_W[i*2+0];
int dstIndex = i + vertexOffset - tableOffset;
clear(vertex, dstIndex, vertexDesc);
clear(varying, dstIndex, varyingDesc);
2013-06-05 17:55:24 +00:00
addWithWeight(vertex, dstIndex, eidx0, vertWeight, vertexDesc);
addWithWeight(vertex, dstIndex, eidx1, vertWeight, vertexDesc);
2013-06-05 17:55:24 +00:00
if (eidx2 != -1) {
float faceWeight = E_W[i*2+1];
addWithWeight(vertex, dstIndex, eidx2, faceWeight, vertexDesc);
addWithWeight(vertex, dstIndex, eidx3, faceWeight, vertexDesc);
2013-06-05 17:55:24 +00:00
}
addWithWeight(varying, dstIndex, eidx0, 0.5f, varyingDesc);
addWithWeight(varying, dstIndex, eidx1, 0.5f, varyingDesc);
}
}
}
2013-06-05 17:55:24 +00:00
TBBEdgeKernel(TBBEdgeKernel const &other)
2013-06-05 17:55:24 +00:00
{
this->vertex = other.vertex;
this->varying= other.varying;
this->vertexDesc = other.vertexDesc;
this->varyingDesc = other.varyingDesc;
2013-06-05 17:55:24 +00:00
this->E_IT = other.E_IT;
this->E_W = other.E_W;
this->vertexOffset = other.vertexOffset;
this->tableOffset = other.tableOffset;
}
TBBEdgeKernel(float *vertex_in,
2013-06-05 17:55:24 +00:00
float *varying_in,
OsdVertexBufferDescriptor const &vertexDesc_in,
OsdVertexBufferDescriptor const &varyingDesc_in,
int const *E_IT_in,
float const *E_W_in,
2013-06-05 17:55:24 +00:00
int vertexOffset_in,
int tableOffset_in) :
vertex (vertex_in),
varying(varying_in),
vertexDesc(vertexDesc_in),
varyingDesc(varyingDesc_in),
2013-06-05 17:55:24 +00:00
E_IT (E_IT_in),
E_W (E_W_in),
vertexOffset(vertexOffset_in),
tableOffset(tableOffset_in)
{};
2013-06-05 17:55:24 +00:00
};
void OsdTbbComputeEdge(
float *vertex, float *varying,
OsdVertexBufferDescriptor const &vertexDesc,
OsdVertexBufferDescriptor const &varyingDesc,
int const *E_IT, float const *E_W, int vertexOffset, int tableOffset,
2013-06-05 17:55:24 +00:00
int start, int end) {
tbb::blocked_range<int> range(start, end, grain_size);
TBBEdgeKernel kernel(vertex, varying, vertexDesc, varyingDesc, E_IT, E_W,
vertexOffset, tableOffset);
2013-06-05 17:55:24 +00:00
tbb::parallel_for(range, kernel);
}
class TBBVertexKernelA {
float *vertex;
float *varying;
OsdVertexBufferDescriptor vertexDesc;
OsdVertexBufferDescriptor varyingDesc;
int const *V_ITa;
float const *V_W;
2013-06-05 17:55:24 +00:00
int vertexOffset;
int tableOffset;
int pass;
public:
void operator() (tbb::blocked_range<int> const &r) const {
if(vertexDesc.length == 4 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeVertexAKernel<4>(vertex, V_ITa, V_W, vertexOffset, tableOffset,
r.begin(), r.end(), pass);
}
else if (vertexDesc.length == 8 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeVertexAKernel<8>(vertex, V_ITa, V_W, vertexOffset, tableOffset,
r.begin(), r.end(), pass);
}
2013-06-05 17:55:24 +00:00
else {
for (int i = r.begin() + tableOffset; i < r.end() + tableOffset; i++) {
int n = V_ITa[5*i+1];
int p = V_ITa[5*i+2];
int eidx0 = V_ITa[5*i+3];
int eidx1 = V_ITa[5*i+4];
float weight = (pass == 1) ? V_W[i] : 1.0f - V_W[i];
// In the case of fractional weight, the weight must be inverted since
// the value is shared with the k_Smooth kernel (statistically the
// k_Smooth kernel runs much more often than this one)
if (weight > 0.0f && weight < 1.0f && n > 0)
weight = 1.0f - weight;
int dstIndex = i + vertexOffset - tableOffset;
if (not pass) {
clear(vertex, dstIndex, vertexDesc);
clear(varying, dstIndex, varyingDesc);
}
2013-06-05 17:55:24 +00:00
if (eidx0 == -1 || (pass == 0 && (n == -1))) {
addWithWeight(vertex, dstIndex, p, weight, vertexDesc);
2013-06-05 17:55:24 +00:00
} else {
addWithWeight(vertex, dstIndex, p, weight * 0.75f, vertexDesc);
addWithWeight(vertex, dstIndex, eidx0, weight * 0.125f, vertexDesc);
addWithWeight(vertex, dstIndex, eidx1, weight * 0.125f, vertexDesc);
2013-06-05 17:55:24 +00:00
}
2013-06-05 17:55:24 +00:00
if (not pass)
addWithWeight(varying, dstIndex, p, 1.0f, varyingDesc);
}
}
}
2013-06-05 17:55:24 +00:00
TBBVertexKernelA(TBBVertexKernelA const &other)
2013-06-05 17:55:24 +00:00
{
this->vertex = other.vertex;
this->varying= other.varying;
this->vertexDesc = other.vertexDesc;
this->varyingDesc = other.varyingDesc;
2013-06-05 17:55:24 +00:00
this->V_ITa = other.V_ITa;
this->V_W = other.V_W;
this->vertexOffset = other.vertexOffset;
this->tableOffset = other.tableOffset;
this->pass = other.pass;
}
TBBVertexKernelA(float *vertex_in,
2013-06-05 17:55:24 +00:00
float *varying_in,
OsdVertexBufferDescriptor const &vertexDesc_in,
OsdVertexBufferDescriptor const &varyingDesc_in,
int const *V_ITa_in,
float const *V_W_in,
2013-06-05 17:55:24 +00:00
int vertexOffset_in,
int tableOffset_in,
int pass_in) :
vertex (vertex_in),
varying(varying_in),
vertexDesc(vertexDesc_in),
varyingDesc(varyingDesc_in),
2013-06-05 17:55:24 +00:00
V_ITa (V_ITa_in),
V_W (V_W_in),
vertexOffset(vertexOffset_in),
tableOffset(tableOffset_in),
pass(pass_in)
{};
2013-06-05 17:55:24 +00:00
};
void OsdTbbComputeVertexA(
float *vertex, float *varying,
OsdVertexBufferDescriptor const &vertexDesc,
OsdVertexBufferDescriptor const &varyingDesc,
int const *V_ITa, float const *V_W, int vertexOffset, int tableOffset,
2013-06-05 17:55:24 +00:00
int start, int end, int pass) {
tbb::blocked_range<int> range(start, end, grain_size);
TBBVertexKernelA kernel(vertex, varying, vertexDesc, varyingDesc,
V_ITa, V_W,
vertexOffset, tableOffset, pass);
2013-06-05 17:55:24 +00:00
tbb::parallel_for(range, kernel);
}
class TBBVertexKernelB {
float *vertex;
float *varying;
OsdVertexBufferDescriptor vertexDesc;
OsdVertexBufferDescriptor varyingDesc;
int const *V_ITa;
int const *V_IT;
float const *V_W;
2013-06-05 17:55:24 +00:00
int vertexOffset;
int tableOffset;
public:
void operator() (tbb::blocked_range<int> const &r) const {
if(vertexDesc.length == 4 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeVertexBKernel<4>(vertex, V_ITa, V_IT, V_W,
vertexOffset, tableOffset, r.begin(), r.end());
}
else if(vertexDesc.length == 8 && varying == NULL) {
2013-06-05 17:55:24 +00:00
ComputeVertexBKernel<8>(vertex, V_ITa, V_IT, V_W,
vertexOffset, tableOffset, r.begin(), r.end());
}
2013-06-05 17:55:24 +00:00
else {
for (int i = r.begin() + tableOffset; i < r.end() + tableOffset; i++) {
int h = V_ITa[5*i];
int n = V_ITa[5*i+1];
int p = V_ITa[5*i+2];
float weight = V_W[i];
float wp = 1.0f/static_cast<float>(n*n);
float wv = (n-2.0f) * n * wp;
int dstIndex = i + vertexOffset - tableOffset;
clear(vertex, dstIndex, vertexDesc);
clear(varying, dstIndex, varyingDesc);
2013-06-05 17:55:24 +00:00
addWithWeight(vertex, dstIndex, p, weight * wv, vertexDesc);
2013-06-05 17:55:24 +00:00
for (int j = 0; j < n; ++j) {
addWithWeight(vertex, dstIndex, V_IT[h+j*2], weight * wp, vertexDesc);
addWithWeight(vertex, dstIndex, V_IT[h+j*2+1], weight * wp, vertexDesc);
2013-06-05 17:55:24 +00:00
}
addWithWeight(varying, dstIndex, p, 1.0f, varyingDesc);
2013-06-05 17:55:24 +00:00
}
}
}
2013-06-05 17:55:24 +00:00
TBBVertexKernelB(TBBVertexKernelB const &other)
2013-06-05 17:55:24 +00:00
{
this->vertex = other.vertex;
this->varying= other.varying;
this->vertexDesc = other.vertexDesc;
this->varyingDesc = other.varyingDesc;
2013-06-05 17:55:24 +00:00
this->V_ITa = other.V_ITa;
this->V_IT = other.V_IT;
this->V_W = other.V_W;
this->vertexOffset = other.vertexOffset;
this->tableOffset = other.tableOffset;
}
TBBVertexKernelB(float *vertex_in,
2013-06-05 17:55:24 +00:00
float *varying_in,
OsdVertexBufferDescriptor const &vertexDesc_in,
OsdVertexBufferDescriptor const &varyingDesc_in,
int const *V_ITa_in,
int const *V_IT_in,
float const *V_W_in,
2013-06-05 17:55:24 +00:00
int vertexOffset_in,
int tableOffset_in) :
vertex (vertex_in),
varying(varying_in),
vertexDesc(vertexDesc_in),
varyingDesc(varyingDesc_in),
2013-06-05 17:55:24 +00:00
V_ITa (V_ITa_in),
V_IT (V_IT_in),
2013-06-05 17:55:24 +00:00
V_W (V_W_in),
vertexOffset(vertexOffset_in),
tableOffset(tableOffset_in)
{};
2013-06-05 17:55:24 +00:00
};
void OsdTbbComputeVertexB(
float *vertex, float *varying,
OsdVertexBufferDescriptor const &vertexDesc,
OsdVertexBufferDescriptor const &varyingDesc,
int const *V_ITa, int const *V_IT, float const *V_W,
2013-06-05 17:55:24 +00:00
int vertexOffset, int tableOffset, int start, int end) {
tbb::blocked_range<int> range(start, end, grain_size);
TBBVertexKernelB kernel(vertex, varying, vertexDesc, varyingDesc,
V_ITa, V_IT, V_W,
vertexOffset, tableOffset);
2013-06-05 17:55:24 +00:00
tbb::parallel_for(range, kernel);
}
class TBBLoopVertexKernelB {
float *vertex;
float *varying;
OsdVertexBufferDescriptor vertexDesc;
OsdVertexBufferDescriptor varyingDesc;
int const *V_ITa;
int const *V_IT;
float const *V_W;
2013-06-05 17:55:24 +00:00
int vertexOffset;
int tableOffset;
public:
void operator() (tbb::blocked_range<int> const &r) const {
if(vertexDesc.length == 4 && varying == NULL) {
ComputeLoopVertexBKernel<4>(vertex, V_ITa, V_IT, V_W, vertexOffset,
2013-06-05 17:55:24 +00:00
tableOffset, r.begin(), r.end());
}
else if(vertexDesc.length == 8 && varying == NULL) {
ComputeLoopVertexBKernel<8>(vertex, V_ITa, V_IT, V_W, vertexOffset,
tableOffset, r.begin(), r.end());
}
2013-06-05 17:55:24 +00:00
else {
for (int i = r.begin() + tableOffset; i < r.end() + tableOffset; i++) {
int h = V_ITa[5*i];
int n = V_ITa[5*i+1];
int p = V_ITa[5*i+2];
float weight = V_W[i];
float wp = 1.0f/static_cast<float>(n);
float beta = 0.25f * cosf(static_cast<float>(M_PI) * 2.0f * wp) + 0.375f;
beta = beta * beta;
beta = (0.625f - beta) * wp;
int dstIndex = i + vertexOffset - tableOffset;
clear(vertex, dstIndex, vertexDesc);
clear(varying, dstIndex, varyingDesc);
2013-06-05 17:55:24 +00:00
addWithWeight(vertex, dstIndex, p, weight * (1.0f - (beta * n)), vertexDesc);
2013-06-05 17:55:24 +00:00
for (int j = 0; j < n; ++j)
addWithWeight(vertex, dstIndex, V_IT[h+j], weight * beta, vertexDesc);
2013-06-05 17:55:24 +00:00
addWithWeight(varying, dstIndex, p, 1.0f, varyingDesc);
}
}
}
2013-06-05 17:55:24 +00:00
TBBLoopVertexKernelB(TBBLoopVertexKernelB const &other)
2013-06-05 17:55:24 +00:00
{
this->vertex = other.vertex;
this->varying= other.varying;
this->vertexDesc = other.vertexDesc;
this->varyingDesc = other.varyingDesc;
2013-06-05 17:55:24 +00:00
this->V_ITa = other.V_ITa;
this->V_IT = other.V_IT;
this->V_W = other.V_W;
this->vertexOffset = other.vertexOffset;
this->tableOffset = other.tableOffset;
}
TBBLoopVertexKernelB(float *vertex_in,
float *varying_in,
OsdVertexBufferDescriptor const &vertexDesc_in,
OsdVertexBufferDescriptor const &varyingDesc_in,
int const *V_ITa_in,
int const *V_IT_in,
float const *V_W_in,
int vertexOffset_in,
int tableOffset_in) :
vertex (vertex_in),
varying(varying_in),
vertexDesc(vertexDesc_in),
varyingDesc(varyingDesc_in),
V_ITa (V_ITa_in),
V_IT (V_IT_in),
V_W (V_W_in),
vertexOffset(vertexOffset_in),
tableOffset(tableOffset_in)
{};
2013-06-05 17:55:24 +00:00
};
void OsdTbbComputeLoopVertexB(
float *vertex, float *varying,
OsdVertexBufferDescriptor const &vertexDesc,
OsdVertexBufferDescriptor const &varyingDesc,
int const *V_ITa, int const *V_IT, float const *V_W,
2013-06-05 17:55:24 +00:00
int vertexOffset, int tableOffset, int start, int end) {
tbb::blocked_range<int> range(start, end, grain_size);
TBBLoopVertexKernelB kernel(vertex, varying, vertexDesc, varyingDesc,
V_ITa, V_IT, V_W,
vertexOffset, tableOffset);
2013-06-05 17:55:24 +00:00
tbb::parallel_for(range, kernel);
}
class TBBBilinearEdgeKernel {
float *vertex;
float *varying;
OsdVertexBufferDescriptor vertexDesc;
OsdVertexBufferDescriptor varyingDesc;
int const *E_IT;
2013-06-05 17:55:24 +00:00
int vertexOffset;
int tableOffset;
public:
void operator() (tbb::blocked_range<int> const &r) const {
if(vertexDesc.length == 4 && varying == NULL) {
ComputeBilinearEdgeKernel<4>(vertex, E_IT, vertexOffset, tableOffset,
2013-06-05 17:55:24 +00:00
r.begin(), r.end());
}
else if(vertexDesc.length == 8 && varying == NULL) {
ComputeBilinearEdgeKernel<8>(vertex, E_IT, vertexOffset, tableOffset,
r.begin(), r.end());
2013-06-05 17:55:24 +00:00
}
else {
for (int i = r.begin() + tableOffset; i < r.end() + tableOffset; i++) {
int eidx0 = E_IT[2*i+0];
int eidx1 = E_IT[2*i+1];
int dstIndex = i + vertexOffset - tableOffset;
clear(vertex, dstIndex, vertexDesc);
clear(varying, dstIndex, varyingDesc);
2013-06-05 17:55:24 +00:00
addWithWeight(vertex, dstIndex, eidx0, 0.5f, vertexDesc);
addWithWeight(vertex, dstIndex, eidx1, 0.5f, vertexDesc);
2013-06-05 17:55:24 +00:00
addWithWeight(varying, dstIndex, eidx0, 0.5f, varyingDesc);
addWithWeight(varying, dstIndex, eidx1, 0.5f, varyingDesc);
}
}
}
2013-06-05 17:55:24 +00:00
TBBBilinearEdgeKernel(TBBBilinearEdgeKernel const &other)
2013-06-05 17:55:24 +00:00
{
this->vertex = other.vertex;
this->varying= other.varying;
this->vertexDesc = other.vertexDesc;
this->varyingDesc = other.varyingDesc;
2013-06-05 17:55:24 +00:00
this->E_IT = other.E_IT;
this->vertexOffset = other.vertexOffset;
this->tableOffset = other.tableOffset;
}
TBBBilinearEdgeKernel(float *vertex_in,
float *varying_in,
OsdVertexBufferDescriptor const &vertexDesc_in,
OsdVertexBufferDescriptor const &varyingDesc_in,
int const *E_IT_in,
int vertexOffset_in,
int tableOffset_in) :
vertex (vertex_in),
varying(varying_in),
vertexDesc(vertexDesc_in),
varyingDesc(varyingDesc_in),
E_IT (E_IT_in),
vertexOffset(vertexOffset_in),
tableOffset(tableOffset_in)
{};
2013-06-05 17:55:24 +00:00
};
void OsdTbbComputeBilinearEdge(
float *vertex, float *varying,
OsdVertexBufferDescriptor const &vertexDesc,
OsdVertexBufferDescriptor const &varyingDesc,
int const *E_IT, int vertexOffset, int tableOffset, int start, int end) {
tbb::blocked_range<int> range(start, end, grain_size);
TBBBilinearEdgeKernel kernel(vertex, varying, vertexDesc, varyingDesc,
E_IT, vertexOffset, tableOffset);
tbb::parallel_for(range, kernel);
2013-06-05 17:55:24 +00:00
}
class TBBBilinearVertexKernel {
float *vertex;
float *varying;
OsdVertexBufferDescriptor vertexDesc;
OsdVertexBufferDescriptor varyingDesc;
int const *V_ITa;
2013-06-05 17:55:24 +00:00
int vertexOffset;
int tableOffset;
public:
void operator() (tbb::blocked_range<int> const &r) const {
int numVertexElements = vertexDesc.length;
int numVaryingElements = varyingDesc.length;
float *src, *des;
2013-06-05 17:55:24 +00:00
for (int i = r.begin() + tableOffset; i < r.end() + tableOffset; i++) {
int p = V_ITa[i];
int dstIndex = i + vertexOffset - tableOffset;
2013-06-05 17:55:24 +00:00
src = vertex + p * numVertexElements;
des = vertex + dstIndex * numVertexElements;
2013-06-05 17:55:24 +00:00
memcpy(des, src, sizeof(float)*numVertexElements);
if(varying) {
src = varying + p * numVaryingElements;
des = varying + dstIndex * numVaryingElements;
2013-06-05 17:55:24 +00:00
memcpy(des, src, sizeof(float)*numVaryingElements);
}
}
}
2013-06-05 17:55:24 +00:00
TBBBilinearVertexKernel(TBBBilinearVertexKernel const &other)
2013-06-05 17:55:24 +00:00
{
this->vertex = other.vertex;
this->varying= other.varying;
this->vertexDesc = other.vertexDesc;
this->varyingDesc = other.varyingDesc;
2013-06-05 17:55:24 +00:00
this->V_ITa = other.V_ITa;
this->vertexOffset = other.vertexOffset;
this->tableOffset = other.tableOffset;
}
TBBBilinearVertexKernel(float *vertex_in,
2013-06-05 17:55:24 +00:00
float *varying_in,
OsdVertexBufferDescriptor const &vertexDesc_in,
OsdVertexBufferDescriptor const &varyingDesc_in,
int const *V_ITa_in,
2013-06-05 17:55:24 +00:00
int vertexOffset_in,
int tableOffset_in) :
vertex (vertex_in),
varying(varying_in),
vertexDesc(vertexDesc_in),
varyingDesc(varyingDesc_in),
V_ITa (V_ITa_in),
2013-06-05 17:55:24 +00:00
vertexOffset(vertexOffset_in),
tableOffset(tableOffset_in)
{};
2013-06-05 17:55:24 +00:00
};
void OsdTbbComputeBilinearVertex(
float *vertex, float *varying,
OsdVertexBufferDescriptor const &vertexDesc,
OsdVertexBufferDescriptor const &varyingDesc,
int const *V_ITa, int vertexOffset, int tableOffset, int start, int end) {
tbb::blocked_range<int> range(start, end, grain_size);
TBBBilinearVertexKernel kernel(vertex, varying, vertexDesc, varyingDesc,
V_ITa, vertexOffset, tableOffset);
tbb::parallel_for(range, kernel);
2013-06-05 17:55:24 +00:00
}
void OsdTbbEditVertexAdd(
float *vertex,
OsdVertexBufferDescriptor const &vertexDesc,
2013-06-05 17:55:24 +00:00
int primVarOffset, int primVarWidth, int vertexOffset, int tableOffset,
int start, int end,
unsigned int const *editIndices, float const *editValues) {
2013-06-05 17:55:24 +00:00
for (int i = start+tableOffset; i < end+tableOffset; i++) {
if (vertex) {
int editIndex = editIndices[i] + vertexOffset;
float *dst = vertex + editIndex * vertexDesc.stride + primVarOffset;
for (int i = 0; i < primVarWidth; ++i) {
dst[i] += editValues[i];
}
}
}
2013-06-05 17:55:24 +00:00
}
void OsdTbbEditVertexSet(
float *vertex,
OsdVertexBufferDescriptor const &vertexDesc,
2013-06-05 17:55:24 +00:00
int primVarOffset, int primVarWidth, int vertexOffset, int tableOffset,
int start, int end,
unsigned int const *editIndices, float const *editValues) {
2013-06-05 17:55:24 +00:00
for (int i = start+tableOffset; i < end+tableOffset; i++) {
if (vertex) {
int editIndex = editIndices[i] + vertexOffset;
float *dst = vertex + editIndex * vertexDesc.stride + primVarOffset;
for (int i = 0; i < primVarWidth; ++i) {
dst[i] = editValues[i];
}
}
2013-06-05 17:55:24 +00:00
}
}
} // end namespace OPENSUBDIV_VERSION
} // end namespace OpenSubdiv