OptimisedUtilProfiler(void)
{
mOptimisedUtils.push_back(_getOptimisedUtilGeneral());
#if __OGRE_HAVE_SSE
if (PlatformInformation::getCpuFeatures() & PlatformInformation::CPU_FEATURE_SSE)
{
mOptimisedUtils.push_back(_getOptimisedUtilSSE());
}
#endif
}
/// @copydoc OptimisedUtil::calculateLightFacing
virtual void calculateLightFacing(
const Vector4& lightPos,
const Vector4* faceNormals,
char* lightFacings,
size_t numFaces)
{
static ProfileItems results;
static size_t index;
index = Root::getSingleton().getNextFrameNumber() % mOptimisedUtils.size();
OptimisedUtil* impl = mOptimisedUtils[index];
ProfileItem& profile = results[index];
profile.begin();
impl->calculateLightFacing(
lightPos,
faceNormals,
lightFacings,
numFaces);
profile.end();
//
// Dagon SkeletonAnimation sample test results (CPU timestamp per-function call):
//
// Pentium 4 3.0G HT Athlon XP 2500+
//
// General 171875 86998
// SSE 47934 63995
//
// You can put break point here while running test application, to
// watch profile results.
++index; // So we can put break point here even if in release build
}
/// @copydoc OptimisedUtil::calculateFaceNormals
virtual void calculateFaceNormals(
const float *positions,
const EdgeData::Triangle *triangles,
Vector4 *faceNormals,
size_t numTriangles)
{
static ProfileItems results;
static size_t index;
index = Root::getSingleton().getNextFrameNumber() % mOptimisedUtils.size();
OptimisedUtil* impl = mOptimisedUtils[index];
ProfileItem& profile = results[index];
profile.begin();
impl->calculateFaceNormals(
positions,
triangles,
faceNormals,
numTriangles);
profile.end();
//
// Dagon SkeletonAnimation sample test results (CPU timestamp per-function call):
//
// Pentium 4 3.0G HT Athlon XP 2500+
//
// General 657080 486494
// SSE 223559 399495
//
// You can put break point here while running test application, to
// watch profile results.
++index; // So we can put break point here even if in release build
}
virtual void softwareVertexMorph(
Real t,
const float *srcPos1, const float *srcPos2,
float *dstPos,
size_t pos1VSize, size_t pos2VSize, size_t dstVSize,
size_t numVertices,
bool morphNormals)
{
static ProfileItems results;
static size_t index;
index = Root::getSingleton().getNextFrameNumber() % mOptimisedUtils.size();
OptimisedUtil* impl = mOptimisedUtils[index];
ProfileItem& profile = results[index];
profile.begin();
impl->softwareVertexMorph(
t,
srcPos1, srcPos2,
dstPos,
pos1VSize, pos2VSize, dstVSize,
numVertices,
morphNormals);
profile.end();
// You can put break point here while running test application, to
// watch profile results.
++index; // So we can put break point here even if in release build
}
virtual void softwareVertexSkinning(
const float *srcPosPtr, float *destPosPtr,
const float *srcNormPtr, float *destNormPtr,
const float *blendWeightPtr, const unsigned char* blendIndexPtr,
const Matrix4* const* blendMatrices,
size_t srcPosStride, size_t destPosStride,
size_t srcNormStride, size_t destNormStride,
size_t blendWeightStride, size_t blendIndexStride,
size_t numWeightsPerVertex,
size_t numVertices)
{
static ProfileItems results;
static size_t index;
index = Root::getSingleton().getNextFrameNumber() % mOptimisedUtils.size();
OptimisedUtil* impl = mOptimisedUtils[index];
ProfileItem& profile = results[index];
profile.begin();
impl->softwareVertexSkinning(
srcPosPtr, destPosPtr,
srcNormPtr, destNormPtr,
blendWeightPtr, blendIndexPtr,
blendMatrices,
srcPosStride, destPosStride,
srcNormStride, destNormStride,
blendWeightStride, blendIndexStride,
numWeightsPerVertex,
numVertices);
profile.end();
// You can put break point here while running test application, to
// watch profile results.
++index; // So we can put break point here even if in release build
}
OptimisedUtilProfiler(void)
{
mOptimisedUtils.push_back(_getOptimisedUtilGeneral());
#if __OGRE_HAVE_SSE
if (PlatformInformation::getCpuFeatures() & PlatformInformation::CPU_FEATURE_SSE)
{
mOptimisedUtils.push_back(_getOptimisedUtilSSE());
}
//#elif __OGRE_HAVE_VFP
// if (PlatformInformation::getCpuFeatures() & PlatformInformation::CPU_FEATURE_VFP)
// {
// mOptimisedUtils.push_back(_getOptimisedUtilVFP());
// }
//#elif __OGRE_HAVE_NEON
// if (PlatformInformation::getCpuFeatures() & PlatformInformation::CPU_FEATURE_NEON)
// {
// mOptimisedUtils.push_back(_getOptimisedUtilNEON());
// }
#endif
}
virtual void extrudeVertices(
const Vector4& lightPos,
Real extrudeDist,
const float* srcPositions,
float* destPositions,
size_t numVertices)
{
static ProfileItems results;
static size_t index;
index = Root::getSingleton().getNextFrameNumber() % mOptimisedUtils.size();
OptimisedUtil* impl = mOptimisedUtils[index];
ProfileItem& profile = results[index];
profile.begin();
impl->extrudeVertices(
lightPos,
extrudeDist,
srcPositions,
destPositions,
numVertices);
profile.end();
//
// Dagon SkeletonAnimation sample test results (CPU timestamp per-function call):
//
// Pentium 4 3.0G HT Athlon XP 2500+
//
// Directional Light, General 38106 92306
// Directional Light, SSE 27292 67055
//
// Point Light, General 224209 155483
// Point Light, SSE 56817 106663
//
// You can put break point here while running test application, to
// watch profile results.
++index; // So we can put break point here even if in release build
}
virtual void concatenateAffineMatrices(
const Matrix4& baseMatrix,
const Matrix4* srcMatrices,
Matrix4* dstMatrices,
size_t numMatrices)
{
static ProfileItems results;
static size_t index;
index = Root::getSingleton().getNextFrameNumber() % mOptimisedUtils.size();
OptimisedUtil* impl = mOptimisedUtils[index];
ProfileItem& profile = results[index];
profile.begin();
impl->concatenateAffineMatrices(
baseMatrix,
srcMatrices,
dstMatrices,
numMatrices);
profile.end();
// You can put break point here while running test application, to
// watch profile results.
++index; // So we can put break point here even if in release build
}
