blender/source/gameengine/Rasterizer/RAS_BucketManager.cpp
Benoit Bolsee 386122ada6 BGE performance, 4th round: logic
This commit extends the technique of dynamic linked list to the logic
system to eliminate as much as possible temporaries, map lookup or 
full scan. The logic engine is now free of memory allocation, which is
an important stability factor. 

The overhead of the logic system is reduced by a factor between 3 and 6
depending on the logic setup. This is the speed-up you can expect on 
a logic setup using simple bricks. Heavy bricks like python controllers
and ray sensors will still take about the same time to execute so the
speed up will be less important.

The core of the logic engine has been much reworked but the functionality
is still the same except for one thing: the priority system on the 
execution of controllers. The exact same remark applies to actuators but
I'll explain for controllers only:

Previously, it was possible, with the "executePriority" attribute to set
a controller to run before any other controllers in the game. Other than
that, the sequential execution of controllers, as defined in Blender was
guaranteed by default.

With the new system, the sequential execution of controllers is still 
guaranteed but only within the controllers of one object. the user can
no longer set a controller to run before any other controllers in the
game. The "executePriority" attribute controls the execution of controllers
within one object. The priority is a small number starting from 0 for the
first controller and incrementing for each controller.

If this missing feature is a must, a special method can be implemented
to set a controller to run before all other controllers.

Other improvements:
- Systematic use of reference in parameter passing to avoid unnecessary data copy
- Use pre increment in iterator instead of post increment to avoid temporary allocation
- Use const char* instead of STR_String whenever possible to avoid temporary allocation
- Fix reference counting bugs (memory leak)
- Fix a crash in certain cases of state switching and object deletion
- Minor speed up in property sensor
- Removal of objects during the game is a lot faster
2009-05-10 20:53:58 +00:00

325 lines
9.0 KiB
C++

/**
* $Id$
* ***** BEGIN GPL LICENSE BLOCK *****
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
* The Original Code is Copyright (C) 2001-2002 by NaN Holding BV.
* All rights reserved.
*
* The Original Code is: all of this file.
*
* Contributor(s): none yet.
*
* ***** END GPL LICENSE BLOCK *****
*/
#ifdef WIN32
// don't show these anoying STL warnings
#pragma warning (disable:4786)
#endif
#include "GEN_Map.h"
#include "RAS_MaterialBucket.h"
#include "STR_HashedString.h"
#include "RAS_MeshObject.h"
#include "RAS_IRasterizer.h"
#include "RAS_IRenderTools.h"
#include "RAS_BucketManager.h"
#include <algorithm>
#include <set>
/* sorting */
struct RAS_BucketManager::sortedmeshslot
{
public:
MT_Scalar m_z; /* depth */
RAS_MeshSlot *m_ms; /* mesh slot */
RAS_MaterialBucket *m_bucket; /* buck mesh slot came from */
sortedmeshslot() {}
void set(RAS_MeshSlot *ms, RAS_MaterialBucket *bucket, const MT_Vector3& pnorm)
{
// would be good to use the actual bounding box center instead
MT_Point3 pos(ms->m_OpenGLMatrix[12], ms->m_OpenGLMatrix[13], ms->m_OpenGLMatrix[14]);
m_z = MT_dot(pnorm, pos);
m_ms = ms;
m_bucket = bucket;
}
};
struct RAS_BucketManager::backtofront
{
bool operator()(const sortedmeshslot &a, const sortedmeshslot &b)
{
return (a.m_z < b.m_z) || (a.m_z == b.m_z && a.m_ms < b.m_ms);
}
};
struct RAS_BucketManager::fronttoback
{
bool operator()(const sortedmeshslot &a, const sortedmeshslot &b)
{
return (a.m_z > b.m_z) || (a.m_z == b.m_z && a.m_ms > b.m_ms);
}
};
/* bucket manager */
RAS_BucketManager::RAS_BucketManager()
{
}
RAS_BucketManager::~RAS_BucketManager()
{
BucketList::iterator it;
for (it = m_SolidBuckets.begin(); it != m_SolidBuckets.end(); it++)
delete (*it);
for (it = m_AlphaBuckets.begin(); it != m_AlphaBuckets.end(); it++)
delete(*it);
m_SolidBuckets.clear();
m_AlphaBuckets.clear();
}
void RAS_BucketManager::OrderBuckets(const MT_Transform& cameratrans, BucketList& buckets, vector<sortedmeshslot>& slots, bool alpha)
{
BucketList::iterator bit;
list<RAS_MeshSlot>::iterator mit;
size_t size = 0, i = 0;
/* Camera's near plane equation: pnorm.dot(point) + pval,
* but we leave out pval since it's constant anyway */
const MT_Vector3 pnorm(cameratrans.getBasis()[2]);
for (bit = buckets.begin(); bit != buckets.end(); ++bit)
{
SG_DList::iterator<RAS_MeshSlot> mit((*bit)->GetActiveMeshSlots());
for(mit.begin(); !mit.end(); ++mit)
size++;
}
slots.resize(size);
for (bit = buckets.begin(); bit != buckets.end(); ++bit)
{
RAS_MaterialBucket* bucket = *bit;
RAS_MeshSlot* ms;
// remove the mesh slot form the list, it culls them automatically for next frame
for(ms = bucket->GetNextActiveMeshSlot();
ms!= NULL;
ms = bucket->GetNextActiveMeshSlot())
{
slots[i++].set(ms, bucket, pnorm);
}
}
if(alpha)
sort(slots.begin(), slots.end(), backtofront());
else
sort(slots.begin(), slots.end(), fronttoback());
}
void RAS_BucketManager::RenderAlphaBuckets(
const MT_Transform& cameratrans, RAS_IRasterizer* rasty, RAS_IRenderTools* rendertools)
{
vector<sortedmeshslot> slots;
vector<sortedmeshslot>::iterator sit;
// Having depth masks disabled/enabled gives different artifacts in
// case no sorting is done or is done inexact. For compatibility, we
// disable it.
rasty->SetDepthMask(RAS_IRasterizer::KX_DEPTHMASK_DISABLED);
OrderBuckets(cameratrans, m_AlphaBuckets, slots, true);
for(sit=slots.begin(); sit!=slots.end(); ++sit) {
rendertools->SetClientObject(rasty, sit->m_ms->m_clientObj);
while(sit->m_bucket->ActivateMaterial(cameratrans, rasty, rendertools))
sit->m_bucket->RenderMeshSlot(cameratrans, rasty, rendertools, *(sit->m_ms));
// make this mesh slot culled automatically for next frame
// it will be culled out by frustrum culling
sit->m_ms->SetCulled(true);
}
rasty->SetDepthMask(RAS_IRasterizer::KX_DEPTHMASK_ENABLED);
}
void RAS_BucketManager::RenderSolidBuckets(
const MT_Transform& cameratrans, RAS_IRasterizer* rasty, RAS_IRenderTools* rendertools)
{
BucketList::iterator bit;
rasty->SetDepthMask(RAS_IRasterizer::KX_DEPTHMASK_ENABLED);
for (bit = m_SolidBuckets.begin(); bit != m_SolidBuckets.end(); ++bit) {
#if 1
RAS_MaterialBucket* bucket = *bit;
RAS_MeshSlot* ms;
// remove the mesh slot form the list, it culls them automatically for next frame
for(ms = bucket->GetNextActiveMeshSlot();
ms!= NULL;
ms = bucket->GetNextActiveMeshSlot())
{
rendertools->SetClientObject(rasty, ms->m_clientObj);
while (bucket->ActivateMaterial(cameratrans, rasty, rendertools))
bucket->RenderMeshSlot(cameratrans, rasty, rendertools, *ms);
// make this mesh slot culled automatically for next frame
// it will be culled out by frustrum culling
ms->SetCulled(true);
}
#else
list<RAS_MeshSlot>::iterator mit;
for (mit = (*bit)->msBegin(); mit != (*bit)->msEnd(); ++mit) {
if (mit->IsCulled())
continue;
rendertools->SetClientObject(rasty, mit->m_clientObj);
while ((*bit)->ActivateMaterial(cameratrans, rasty, rendertools))
(*bit)->RenderMeshSlot(cameratrans, rasty, rendertools, *mit);
// make this mesh slot culled automatically for next frame
// it will be culled out by frustrum culling
mit->SetCulled(true);
}
#endif
}
/* this code draws meshes order front-to-back instead to reduce overdraw.
* it turned out slower due to much material state switching, a more clever
* algorithm might do better. */
#if 0
vector<sortedmeshslot> slots;
vector<sortedmeshslot>::iterator sit;
OrderBuckets(cameratrans, m_SolidBuckets, slots, false);
for(sit=slots.begin(); sit!=slots.end(); ++sit) {
rendertools->SetClientObject(rasty, sit->m_ms->m_clientObj);
while(sit->m_bucket->ActivateMaterial(cameratrans, rasty, rendertools))
sit->m_bucket->RenderMeshSlot(cameratrans, rasty, rendertools, *(sit->m_ms));
}
#endif
}
void RAS_BucketManager::Renderbuckets(
const MT_Transform& cameratrans, RAS_IRasterizer* rasty, RAS_IRenderTools* rendertools)
{
/* beginning each frame, clear (texture/material) caching information */
rasty->ClearCachingInfo();
RenderSolidBuckets(cameratrans, rasty, rendertools);
RenderAlphaBuckets(cameratrans, rasty, rendertools);
rendertools->SetClientObject(rasty, NULL);
}
RAS_MaterialBucket* RAS_BucketManager::FindBucket(RAS_IPolyMaterial * material, bool &bucketCreated)
{
BucketList::iterator it;
bucketCreated = false;
for (it = m_SolidBuckets.begin(); it != m_SolidBuckets.end(); it++)
if (*(*it)->GetPolyMaterial() == *material)
return *it;
for (it = m_AlphaBuckets.begin(); it != m_AlphaBuckets.end(); it++)
if (*(*it)->GetPolyMaterial() == *material)
return *it;
RAS_MaterialBucket *bucket = new RAS_MaterialBucket(material);
bucketCreated = true;
if (bucket->IsAlpha())
m_AlphaBuckets.push_back(bucket);
else
m_SolidBuckets.push_back(bucket);
return bucket;
}
void RAS_BucketManager::OptimizeBuckets(MT_Scalar distance)
{
BucketList::iterator bit;
distance = 10.0;
for (bit = m_SolidBuckets.begin(); bit != m_SolidBuckets.end(); ++bit)
(*bit)->Optimize(distance);
for (bit = m_AlphaBuckets.begin(); bit != m_AlphaBuckets.end(); ++bit)
(*bit)->Optimize(distance);
}
void RAS_BucketManager::ReleaseDisplayLists(RAS_IPolyMaterial *mat)
{
BucketList::iterator bit;
list<RAS_MeshSlot>::iterator mit;
for (bit = m_SolidBuckets.begin(); bit != m_SolidBuckets.end(); ++bit) {
if (mat == NULL || (mat == (*bit)->GetPolyMaterial())) {
for (mit = (*bit)->msBegin(); mit != (*bit)->msEnd(); ++mit) {
if(mit->m_DisplayList) {
mit->m_DisplayList->Release();
mit->m_DisplayList = NULL;
}
}
}
}
for (bit = m_AlphaBuckets.begin(); bit != m_AlphaBuckets.end(); ++bit) {
if (mat == NULL || (mat == (*bit)->GetPolyMaterial())) {
for (mit = (*bit)->msBegin(); mit != (*bit)->msEnd(); ++mit) {
if(mit->m_DisplayList) {
mit->m_DisplayList->Release();
mit->m_DisplayList = NULL;
}
}
}
}
}
void RAS_BucketManager::ReleaseMaterials(RAS_IPolyMaterial * mat)
{
BucketList::iterator bit;
list<RAS_MeshSlot>::iterator mit;
for (bit = m_SolidBuckets.begin(); bit != m_SolidBuckets.end(); ++bit) {
if (mat == NULL || (mat == (*bit)->GetPolyMaterial())) {
(*bit)->GetPolyMaterial()->ReleaseMaterial();
}
}
for (bit = m_AlphaBuckets.begin(); bit != m_AlphaBuckets.end(); ++bit) {
if (mat == NULL || (mat == (*bit)->GetPolyMaterial())) {
(*bit)->GetPolyMaterial()->ReleaseMaterial();
}
}
}