ShuriZma
/
dolphin
mirror of https://github.com/dolphin-emu/dolphin.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Massive style & comment cleanup of (mostly) GL plugin - also split some large files. A minor speedup for BP writes - merged the two switch()-es. 

git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@899 8ced0084-cf51-0410-be5f-012b33b47a6e
This commit is contained in:
hrydgard 2008-10-17 11:30:14 +00:00
parent 4477f77cf6
commit dcbc8e78d4
45 changed files with 1288 additions and 1554 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
Source/Core/Common/Src/Common.h
View File

@ -49,10 +49,12 @@
#endif
#ifdef _WIN32
#ifdef _WIN32
// By default, MS' stdio implementation does not support 64-bit offsets.
// This little hack fixes that, keeping the code portable to linux where fseek and fread
// do support 64-bit offsets in modern distributions.
#define fseek _fseeki64
#define ftell _ftelli64
#endif
#define POSIX 0
#define NOMINMAX
@ -128,11 +130,9 @@ typedef union _LARGE_INTEGER
#undef max
template<class T>
inline T min(const T& a, const T& b) {return(a > b ? b : a);}
inline T min(const T& a, const T& b) {return a > b ? b : a;}
template<class T>
inline T max(const T& a, const T& b) {return(a > b ? a : b);}
inline T max(const T& a, const T& b) {return a > b ? a : b;}
// Byte ordering

9
Source/Core/DiscIO/Src/FileSystemGCWii.cpp
View File

@ -158,11 +158,11 @@ bool CFileSystemGCWii::InitFileSystem()
{
if (Read32(0x18) == 0x5D1C9EA3)
{
m_OffsetShift = 2;
m_OffsetShift = 2; // Wii file system
}
else if (Read32(0x1c) == 0xC2339F3D)
{
m_OffsetShift = 0;
m_OffsetShift = 0; // GC file system
}
else
{
@ -206,8 +206,6 @@ bool CFileSystemGCWii::InitFileSystem()
return true;
}
// __________________________________________________________________________________________________
//
// Changed this stuff from C++ string to C strings for speed in debug mode. Doesn't matter in release, but
// std::string is SLOW in debug mode.
size_t CFileSystemGCWii::BuildFilenames(const size_t _FirstIndex, const size_t _LastIndex, const char* _szDirectory, u64 _NameTableOffset)
@ -248,5 +246,4 @@ size_t CFileSystemGCWii::BuildFilenames(const size_t _FirstIndex, const size_t _
return CurrentIndex;
}
} // namespace
} // namespace

10
Source/Core/VideoCommon/Src/DataReader.h
View File

@ -18,11 +18,8 @@
#ifndef _DATAREADER_H
#define _DATAREADER_H
extern u8* g_pVideoData;
inline u8 DataPeek8(u32 _uOffset)
{
return g_pVideoData[_uOffset];
@ -46,14 +43,14 @@ inline u8 DataReadU8()
inline u16 DataReadU16()
{
u16 tmp = Common::swap16(*(u16*)g_pVideoData);
g_pVideoData+=2;
g_pVideoData += 2;
return tmp;
}
inline u32 DataReadU32()
{
u32 tmp = Common::swap32(*(u32*)g_pVideoData);
g_pVideoData+=4;
g_pVideoData += 4;
return tmp;
}
@ -61,7 +58,7 @@ inline float DataReadF32()
{
union {u32 i; float f;} temp;
temp.i = Common::swap32(*(u32*)g_pVideoData);
g_pVideoData+=4;
g_pVideoData += 4;
float tmp = temp.f;
return tmp;
}
@ -77,4 +74,3 @@ inline void DataSkip(u32 skip)
}
#endif

128
Source/Core/VideoCommon/Src/Profiler.cpp
View File

@ -15,20 +15,14 @@
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
#include "Common.h"
// Simple profiler
#include "Common.h"
#include "Profiler.h"
////////////////////
// Small profiler //
////////////////////
#include <list>
#include <string>
#include <map>
using namespace std;
int g_bWriteProfile=0;
#ifdef _WIN32
@ -45,7 +39,14 @@ int g_bWriteProfile=0;
#pragma intrinsic(__rdtsc)
#endif
static u64 luPerfFreq=0;
// Globals
static u64 luPerfFreq = 0;
#ifdef DVPROFILE
int g_bWriteProfile = 1;
#else
int g_bWriteProfile = 1;
#endif
inline u64 GET_PROFILE_TIME()
{
#if defined (_MSC_VER) && _MSC_VER >= 1400
@ -57,11 +58,10 @@ inline u64 GET_PROFILE_TIME()
#endif
}
#else
static u64 luPerfFreq=1000000;
static u64 luPerfFreq = 1000000;
#define GET_PROFILE_TIME() //GetCpuTick()
#endif
struct DVPROFSTRUCT;
struct DVPROFSTRUCT
@ -70,25 +70,27 @@ struct DVPROFSTRUCT
{
DATA(u64 time, u32 user = 0) : dwTime(time), dwUserData(user) {}
DATA() : dwTime(0), dwUserData(0) {}
u64 dwTime;
u32 dwUserData;
};
~DVPROFSTRUCT() {
list<DVPROFSTRUCT*>::iterator it = listpChild.begin();
while(it != listpChild.end() ) {
delete *it; *it = NULL;
std::list<DVPROFSTRUCT *>::iterator it = listpChild.begin();
while (it != listpChild.end()) {
delete *it;
*it = NULL;
++it;
}
}
list<DATA> listTimes; // before DVProfEnd is called, contains the global time it started
// after DVProfEnd is called, contains the time it lasted
// the list contains all the tracked times
// before DVProfEnd is called, contains the global time it started
// after DVProfEnd is called, contains the time it lasted
// the list contains all the tracked times
std::list<DATA> listTimes;
char pname[256];
list<DVPROFSTRUCT*> listpChild; // other profilers called during this profiler period
std::list<DVPROFSTRUCT*> listpChild; // other profilers called during this profiler period
};
struct DVPROFTRACK
@ -98,13 +100,19 @@ struct DVPROFTRACK
DVPROFSTRUCT* pprof;
};
list<DVPROFTRACK> g_listCurTracking; // the current profiling functions, the back element is the
// one that will first get popped off the list when DVProfEnd is called
// the pointer is an element in DVPROFSTRUCT::listTimes
list<DVPROFSTRUCT> g_listProfilers; // the current profilers, note that these are the parents
// any profiler started during the time of another is held in
// DVPROFSTRUCT::listpChild
list<DVPROFSTRUCT*> g_listAllProfilers; // ignores the hierarchy, pointer to elements in g_listProfilers
// the current profiling functions, the back element is the
// one that will first get popped off the list when DVProfEnd is called
// the pointer is an element in DVPROFSTRUCT::listTimes
static std::list<DVPROFTRACK> g_listCurTracking;
// the current profilers, note that these are the parents
// any profiler started during the time of another is held in
// DVPROFSTRUCT::listpChild
static std::list<DVPROFSTRUCT> g_listProfilers;
// ignores the hierarchy, pointer to elements in g_listProfilers
static std::list<DVPROFSTRUCT*> g_listAllProfilers;
void DVProfRegister(const char *pname)
{
@ -123,7 +131,7 @@ void DVProfRegister(const char *pname)
}
#endif
list<DVPROFSTRUCT*>::iterator it = g_listAllProfilers.begin();
std::list<DVPROFSTRUCT*>::iterator it = g_listAllProfilers.begin();
// while(it != g_listAllProfilers.end() ) {
//
@ -191,19 +199,15 @@ struct DVTIMEINFO
u64 uInclusive, uExclusive;
};
map<string, DVTIMEINFO> mapAggregateTimes;
std::map<std::string, DVTIMEINFO> mapAggregateTimes;
u64 DVProfWriteStruct(FILE* f, DVPROFSTRUCT* p, int ident)
u64 DVProfWriteStruct(FILE* f, const DVPROFSTRUCT* p, int ident)
{
fprintf(f, "%*s%s - ", ident, "", p->pname);
list<DVPROFSTRUCT::DATA>::iterator ittime = p->listTimes.begin();
std::list<DVPROFSTRUCT::DATA>::const_iterator ittime = p->listTimes.begin();
u64 utime = 0;
while(ittime != p->listTimes.end() ) {
while (ittime != p->listTimes.end()) {
utime += ittime->dwTime;
if (ittime->dwUserData)
fprintf(f, "time: %d, user: 0x%8.8x", (u32)ittime->dwTime, ittime->dwUserData);
else
@ -212,9 +216,9 @@ u64 DVProfWriteStruct(FILE* f, DVPROFSTRUCT* p, int ident)
}
// yes this is necessary, maps have problems with constructors on their type
map<string, DVTIMEINFO>::iterator ittimes = mapAggregateTimes.find(p->pname);
std::map<std::string, DVTIMEINFO>::iterator ittimes = mapAggregateTimes.find(p->pname);
if (ittimes == mapAggregateTimes.end()) {
ittimes = mapAggregateTimes.insert(map<string, DVTIMEINFO>::value_type(p->pname, DVTIMEINFO())).first;
ittimes = mapAggregateTimes.insert(std::map<std::string, DVTIMEINFO>::value_type(p->pname, DVTIMEINFO())).first;
ittimes->second.uExclusive = 0;
ittimes->second.uInclusive = 0;
}
@ -223,11 +227,10 @@ u64 DVProfWriteStruct(FILE* f, DVPROFSTRUCT* p, int ident)
fprintf(f, "\n");
list<DVPROFSTRUCT*>::iterator itprof = p->listpChild.begin();
std::list<DVPROFSTRUCT*>::const_iterator itprof = p->listpChild.begin();
u64 uex = utime;
while(itprof != p->listpChild.end() ) {
while (itprof != p->listpChild.end()) {
uex -= DVProfWriteStruct(f, *itprof, ident+4);
++itprof;
}
@ -247,38 +250,35 @@ void DVProfWrite(const char* pfilename, u32 frames)
// pop back any unused
mapAggregateTimes.clear();
list<DVPROFSTRUCT>::iterator it = g_listProfilers.begin();
std::list<DVPROFSTRUCT>::iterator it = g_listProfilers.begin();
while(it != g_listProfilers.end() ) {
while (it != g_listProfilers.end() ) {
DVProfWriteStruct(f, &(*it), 0);
++it;
}
{
map<string, DVTIMEINFO>::iterator iter;
fprintf(f, "\n\n-------------------------------------------------------------------\n\n");
std::map<std::string, DVTIMEINFO>::const_iterator iter;
fprintf(f, "\n\n-------------------------------------------------------------------\n\n");
u64 uTotal[2] = {0};
double fiTotalTime[2];
u64 uTotal[2] = {0};
double fiTotalTime[2];
for(iter = mapAggregateTimes.begin(); iter != mapAggregateTimes.end(); ++iter) {
uTotal[0] += iter->second.uExclusive;
uTotal[1] += iter->second.uInclusive;
}
fprintf(f, "total times (%d): ex: %Lu ", frames, 1000000*uTotal[0]/(luPerfFreq*(u64)frames));
fprintf(f, "inc: %Lu\n", 1000000 * uTotal[1]/(luPerfFreq*(u64)frames));
fiTotalTime[0] = 1.0 / (double)uTotal[0];
fiTotalTime[1] = 1.0 / (double)uTotal[1];
// output the combined times
for(iter = mapAggregateTimes.begin(); iter != mapAggregateTimes.end(); ++iter) {
fprintf(f, "%s - ex: %f inc: %f\n", iter->first.c_str(), (float)((double)iter->second.uExclusive * fiTotalTime[0]),
(float)((double)iter->second.uInclusive * fiTotalTime[1]));
}
for (iter = mapAggregateTimes.begin(); iter != mapAggregateTimes.end(); ++iter) {
uTotal[0] += iter->second.uExclusive;
uTotal[1] += iter->second.uInclusive;
}
fprintf(f, "total times (%d): ex: %Lu ", frames, 1000000 * uTotal[0] / (luPerfFreq*(u64)frames));
fprintf(f, "inc: %Lu\n", 1000000 * uTotal[1]/(luPerfFreq*(u64)frames));
fiTotalTime[0] = 1.0 / (double)uTotal[0];
fiTotalTime[1] = 1.0 / (double)uTotal[1];
// output the combined times
for (iter = mapAggregateTimes.begin(); iter != mapAggregateTimes.end(); ++iter) {
fprintf(f, "%s - ex: %f inc: %f\n", iter->first.c_str(), (float)((double)iter->second.uExclusive * fiTotalTime[0]),
(float)((double)iter->second.uInclusive * fiTotalTime[1]));
}
fclose(f);
}

14
Source/Core/VideoCommon/Src/Profiler.h
View File

@ -21,17 +21,23 @@
#ifndef _PROFILER_H
#define _PROFILER_H
#include <string>
#include "Common.h"
// #define DVPROFILE // comment out to disable profiling
extern int g_bWriteProfile; // global variable to enable/disable profiling (if DVPROFILE is defined)
// IMPORTANT: For every Register there must be an End
// IMPORTANT: For every Register there must be an End. Use the below DVProfileFunc utility class for safety.
void DVProfRegister(const char* pname); // first checks if this profiler exists in g_listProfilers
void DVProfEnd(u32 dwUserData);
void DVProfWrite(const char* pfilename, u32 frames = 0);
void DVProfClear(); // clears all the profi lers
#if defined(DVPROFILE) && (defined(_WIN32)||defined(WIN32))
void DVProfWrite(const char* pfilename, u32 frames = 0);
void DVProfGenReport(std::string *report);
void DVProfClear(); // clears all the profilers
#if defined(DVPROFILE) && defined(_WIN32)
#ifdef _MSC_VER

8
Source/Core/VideoCommon/VideoCommon.vcproj
View File

@ -490,6 +490,14 @@
AssemblerOutput="4"
/>
</FileConfiguration>
<FileConfiguration
Name="Release|x64"
>
<Tool
Name="VCCLCompilerTool"
AssemblerOutput="4"
/>
</FileConfiguration>
</File>
<File
RelativePath=".\Src\XFBConvert.h"

21
Source/Plugins/Plugin_DSP_HLE/Src/Globals.h
View File

@ -1,8 +1,23 @@
// Copyright (C) 2003-2008 Dolphin Project.
// 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, version 2.0.
// 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 2.0 for more details.
// A copy of the GPL 2.0 should have been included with the program.
// If not, see http://www.gnu.org/licenses/
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
#ifndef _GLOBALS_H
#define _GLOBALS_H
#include "Common.h"
#include "pluginspecs_dsp.h"
@ -14,6 +29,4 @@ u16 Memory_Read_U16(u32 _uAddress);
u32 Memory_Read_U32(u32 _uAddress);
float Memory_Read_Float(u32 _uAddress);
#endif

16
Source/Plugins/Plugin_DSP_LLE/Plugin_DSP_LLE.vcproj
View File

@ -763,6 +763,22 @@
>
</File>
</Filter>
<Filter
Name="Docs"
>
<File
RelativePath="..\..\..\Docs\DSP\Crazy Taxi.txt"
>
</File>
<File
RelativePath="..\..\..\Docs\DSP\dsp_rom.txt"
>
</File>
<File
RelativePath="..\..\..\Docs\DSP\DSP_UC_Zelda.txt"
>
</File>
</Filter>
<File
RelativePath=".\Src\DSoundStream.cpp"
>

1
Source/Plugins/Plugin_VideoDX9/Src/BPStructs.cpp
View File

@ -19,6 +19,7 @@
#include "Globals.h"
#include "Common.h"
#include "Profiler.h"
#include "BPStructs.h"
#include "OpcodeDecoding.h"
#include "TextureCache.h"

1
Source/Plugins/Plugin_VideoDX9/Src/BPStructs.h
View File

@ -23,7 +23,6 @@
#include "BPMemory.h"
void BPInit();
//bool BPWritten(int addr, int changes);
void LoadBPReg(u32 value0);
void ActivateTextures();

1
Source/Plugins/Plugin_VideoDX9/Src/OpcodeDecoding.cpp
View File

@ -28,6 +28,7 @@
#include "D3DBase.h"
#include "Common.h"
#include "Profiler.h"
#include "Globals.h"
#include "VertexHandler.h"
#include "TransformEngine.h"

1
Source/Plugins/Plugin_VideoDX9/Src/ShaderManager.cpp
View File

@ -17,6 +17,7 @@
#include "D3DBase.h"
#include "Utils.h"
#include "Profiler.h"
#include "Globals.h"
#include "ShaderManager.h"
#include "VertexLoader.h"

1
Source/Plugins/Plugin_VideoDX9/Src/TransformEngine.cpp
View File

@ -18,6 +18,7 @@
#include <fvec.h>
#include "Common.h"
#include "Profiler.h"
#include "Globals.h"
#include "Vec3.h"
#include "TransformEngine.h"

271
Source/Plugins/Plugin_VideoDX9/Src/Utils.cpp
View File

@ -43,274 +43,3 @@ LRESULT CALLBACK AboutProc(HWND hDlg, UINT message, WPARAM wParam, LPARAM /*lPar
}
return FALSE;
}
////////////////////
// Small profiler //
////////////////////
#include <list>
#include <string>
#include <map>
using namespace std;
int g_bWriteProfile=0;
#ifdef _WIN32
#if defined (_MSC_VER) && _MSC_VER >= 1400
#define _interlockedbittestandset workaround_ms_header_bug_platform_sdk6_set
#define _interlockedbittestandreset workaround_ms_header_bug_platform_sdk6_reset
#define _interlockedbittestandset64 workaround_ms_header_bug_platform_sdk6_set64
#define _interlockedbittestandreset64 workaround_ms_header_bug_platform_sdk6_reset64
#include <intrin.h>
#undef _interlockedbittestandset
#undef _interlockedbittestandreset
#undef _interlockedbittestandset64
#undef _interlockedbittestandreset64
#pragma intrinsic(__rdtsc)
#endif
static u64 luPerfFreq=0;
inline u64 GET_PROFILE_TIME()
{
#if defined (_MSC_VER) && _MSC_VER >= 1400
return __rdtsc();
#else
LARGE_INTEGER lu;
QueryPerformanceCounter(&lu);
return lu.QuadPart;
#endif
}
#else
static u64 luPerfFreq = 1000000;
#define GET_PROFILE_TIME() //GetCpuTick()
#endif
struct DVPROFSTRUCT;
struct DVPROFSTRUCT
{
struct DATA
{
DATA(u64 time, u32 user = 0) : dwTime(time), dwUserData(user) {}
DATA() : dwTime(0), dwUserData(0) {}
u64 dwTime;
u32 dwUserData;
};
~DVPROFSTRUCT() {
list<DVPROFSTRUCT*>::iterator it = listpChild.begin();
while(it != listpChild.end() ) {
delete *it; *it = NULL;
++it;
}
}
list<DATA> listTimes; // before DVProfEnd is called, contains the global time it started
// after DVProfEnd is called, contains the time it lasted
// the list contains all the tracked times
char pname[256];
list<DVPROFSTRUCT*> listpChild; // other profilers called during this profiler period
};
struct DVPROFTRACK
{
u32 dwUserData;
DVPROFSTRUCT::DATA* pdwTime;
DVPROFSTRUCT* pprof;
};
list<DVPROFTRACK> g_listCurTracking; // the current profiling functions, the back element is the
// one that will first get popped off the list when DVProfEnd is called
// the pointer is an element in DVPROFSTRUCT::listTimes
list<DVPROFSTRUCT> g_listProfilers; // the current profilers, note that these are the parents
// any profiler started during the time of another is held in
// DVPROFSTRUCT::listpChild
list<DVPROFSTRUCT*> g_listAllProfilers; // ignores the hierarchy, pointer to elements in g_listProfilers
void DVProfRegister(char* pname)
{
if( !g_bWriteProfile )
return;
#ifdef _WIN32
if( luPerfFreq <= 1 ) {
#if defined (_MSC_VER) && _MSC_VER >= 1400
luPerfFreq = 1000000;
#else
LARGE_INTEGER temp;
QueryPerformanceFrequency(&temp);
luPerfFreq = temp.QuadPart;
#endif
}
#endif
list<DVPROFSTRUCT*>::iterator it = g_listAllProfilers.begin();
// while(it != g_listAllProfilers.end() ) {
//
// if( _tcscmp(pname, (*it)->pname) == 0 ) {
// (*it)->listTimes.push_back(timeGetTime());
// DVPROFTRACK dvtrack;
// dvtrack.pdwTime = &(*it)->listTimes.back();
// dvtrack.pprof = *it;
// g_listCurTracking.push_back(dvtrack);
// return;
// }
//
// ++it;
// }
// else add in a new profiler to the appropriate parent profiler
DVPROFSTRUCT* pprof = NULL;
if( g_listCurTracking.size() > 0 ) {
assert( g_listCurTracking.back().pprof != NULL );
g_listCurTracking.back().pprof->listpChild.push_back(new DVPROFSTRUCT());
pprof = g_listCurTracking.back().pprof->listpChild.back();
}
else {
g_listProfilers.push_back(DVPROFSTRUCT());
pprof = &g_listProfilers.back();
}
strncpy(pprof->pname, pname, 256);
// setup the profiler for tracking
pprof->listTimes.push_back(DVPROFSTRUCT::DATA(GET_PROFILE_TIME()));
DVPROFTRACK dvtrack;
dvtrack.pdwTime = &pprof->listTimes.back();
dvtrack.pprof = pprof;
dvtrack.dwUserData = 0;
g_listCurTracking.push_back(dvtrack);
// add to all profiler list
g_listAllProfilers.push_back(pprof);
}
void DVProfEnd(u32 dwUserData)
{
if( !g_bWriteProfile )
return;
if( g_listCurTracking.size() == 0 )
return;
DVPROFTRACK dvtrack = g_listCurTracking.back();
assert( dvtrack.pdwTime != NULL && dvtrack.pprof != NULL );
dvtrack.pdwTime->dwTime = GET_PROFILE_TIME()- dvtrack.pdwTime->dwTime;
dvtrack.pdwTime->dwUserData= dwUserData;
g_listCurTracking.pop_back();
}
struct DVTIMEINFO
{
DVTIMEINFO() : uInclusive(0), uExclusive(0) {}
u64 uInclusive, uExclusive;
};
map<string, DVTIMEINFO> mapAggregateTimes;
u64 DVProfWriteStruct(FILE* f, DVPROFSTRUCT* p, int ident)
{
fprintf(f, "%*s%s - ", ident, "", p->pname);
list<DVPROFSTRUCT::DATA>::iterator ittime = p->listTimes.begin();
u64 utime = 0;
while(ittime != p->listTimes.end() ) {
utime += ittime->dwTime;
if( ittime->dwUserData )
fprintf(f, "time: %d, user: 0x%8.8x", (u32)ittime->dwTime, ittime->dwUserData);
else
fprintf(f, "time: %d", (u32)ittime->dwTime);
++ittime;
}
// yes this is necessary, maps have problems with constructors on their type
map<string, DVTIMEINFO>::iterator ittimes = mapAggregateTimes.find(p->pname);
if( ittimes == mapAggregateTimes.end() ) {
ittimes = mapAggregateTimes.insert(map<string, DVTIMEINFO>::value_type(p->pname, DVTIMEINFO())).first;
ittimes->second.uExclusive = 0;
ittimes->second.uInclusive = 0;
}
ittimes->second.uInclusive += utime;
fprintf(f, "\n");
list<DVPROFSTRUCT*>::iterator itprof = p->listpChild.begin();
u64 uex = utime;
while(itprof != p->listpChild.end() ) {
uex -= DVProfWriteStruct(f, *itprof, ident+4);
++itprof;
}
if( uex > utime ) {
uex = 0;
}
ittimes->second.uExclusive += uex;
return utime;
}
void DVProfWrite(char* pfilename, u32 frames)
{
assert( pfilename != NULL );
FILE* f = fopen(pfilename, "w");
// pop back any unused
mapAggregateTimes.clear();
list<DVPROFSTRUCT>::iterator it = g_listProfilers.begin();
while(it != g_listProfilers.end() ) {
DVProfWriteStruct(f, &(*it), 0);
++it;
}
{
map<string, DVTIMEINFO>::iterator it;
fprintf(f, "\n\n-------------------------------------------------------------------\n\n");
u64 uTotal[2] = {0};
double fiTotalTime[2];
for(it = mapAggregateTimes.begin(); it != mapAggregateTimes.end(); ++it) {
uTotal[0] += it->second.uExclusive;
uTotal[1] += it->second.uInclusive;
}
fprintf(f, "total times (%d): ex: %Lu ", frames, 1000000*uTotal[0]/(luPerfFreq*(u64)frames));
fprintf(f, "inc: %Lu\n", 1000000 * uTotal[1]/(luPerfFreq*(u64)frames));
fiTotalTime[0] = 1.0 / (double)uTotal[0];
fiTotalTime[1] = 1.0 / (double)uTotal[1];
// output the combined times
for(it = mapAggregateTimes.begin(); it != mapAggregateTimes.end(); ++it) {
fprintf(f, "%s - ex: %f inc: %f\n", it->first.c_str(), (float)((double)it->second.uExclusive * fiTotalTime[0]),
(float)((double)it->second.uInclusive * fiTotalTime[1]));
}
}
fclose(f);
}
void DVProfClear()
{
g_listCurTracking.clear();
g_listProfilers.clear();
g_listAllProfilers.clear();
}

50
Source/Plugins/Plugin_VideoDX9/Src/Utils.h
View File

@ -60,54 +60,4 @@ inline float Memory_Read_Float(u32 _uAddress)
return *(float*)&uTemp;
}
////
// profiling
///
extern int g_bWriteProfile; // global variable to enable/disable profiling (if DVPROFILE is defined)
// IMPORTANT: For every Register there must be an End
void DVProfRegister(char* pname); // first checks if this profiler exists in g_listProfilers
void DVProfEnd(u32 dwUserData);
void DVProfWrite(char* pfilename, u32 frames = 0);
void DVProfClear(); // clears all the profilers
// #define DVPROFILE // comment out to disable profiling
#if defined(DVPROFILE) && (defined(_WIN32)||defined(WIN32))
#ifdef _MSC_VER
#ifndef __PRETTY_FUNCTION__
#define __PRETTY_FUNCTION__ __FUNCTION__
#endif
#endif
#define DVSTARTPROFILE() DVProfileFunc _pf(__PRETTY_FUNCTION__);
class DVProfileFunc
{
public:
u32 dwUserData;
DVProfileFunc(char* pname) { DVProfRegister(pname); dwUserData = 0; }
DVProfileFunc(char* pname, u32 dwUserData) : dwUserData(dwUserData) { DVProfRegister(pname); }
~DVProfileFunc() { DVProfEnd(dwUserData); }
};
#else
#define DVSTARTPROFILE()
class DVProfileFunc
{
public:
u32 dwUserData;
__forceinline DVProfileFunc(char* pname) {}
__forceinline DVProfileFunc(char* pname, u32 dwUserData) { }
~DVProfileFunc() {}
};
#endif
#endif

1
Source/Plugins/Plugin_VideoDX9/Src/VertexHandler.cpp
View File

@ -18,6 +18,7 @@
#include "D3DBase.h"
#include "Common.h"
#include "Profiler.h"
#include "Globals.h"
#include "VertexHandler.h"
#include "OpcodeDecoding.h"

1
Source/Plugins/Plugin_VideoDX9/Src/VertexLoader.cpp
View File

@ -20,6 +20,7 @@
#include "x64Emitter.h"
#include "Common.h"
#include "Profiler.h"
#include "VertexHandler.h"
#include "VertexLoader.h"
#include "XFStructs.h"

1
Source/Plugins/Plugin_VideoDX9/Src/XFStructs.cpp
View File

@ -16,6 +16,7 @@
// http://code.google.com/p/dolphin-emu/
#include "stdafx.h"
#include "Profiler.h"
#include "XFStructs.h"
#include "Render.h"
#include "main.h"

18
Source/Plugins/Plugin_VideoOGL/Plugin_VideoOGL.vcproj
View File

@ -1,7 +1,7 @@
<?xml version="1.0" encoding="Windows-1252"?>
<VisualStudioProject
ProjectType="Visual C++"
Version="9.00"
Version="9,00"
Name="Plugin_VideoOGL"
ProjectGUID="{CFDCEE0E-FA45-4F72-9FCC-0B88F5A75160}"
RootNamespace="Plugin_VideoOGL"
@ -760,6 +760,14 @@
RelativePath=".\Src\VertexLoader_TextCoord.h"
>
</File>
<File
RelativePath=".\Src\VertexManager.cpp"
>
</File>
<File
RelativePath=".\Src\VertexManager.h"
>
</File>
</Filter>
<Filter
Name="Render"
@ -1072,6 +1080,14 @@
RelativePath=".\Src\Logging\Console.h"
>
</File>
<File
RelativePath=".\Src\ImageWrite.cpp"
>
</File>
<File
RelativePath=".\Src\ImageWrite.h"
>
</File>
<File
RelativePath=".\Src\Logging\Logging.cpp"
>

410
Source/Plugins/Plugin_VideoOGL/Src/BPStructs.cpp
View File

@ -22,6 +22,7 @@
#include "Profiler.h"
#include "VertexLoader.h"
#include "VertexManager.h"
#include "BPStructs.h"
#include "Render.h"
@ -65,18 +66,13 @@ void BPInit()
}
// =======================================================================================
// Called att the end of every: OpcodeDecoding.cpp ExecuteDisplayList > Decode() >
// LoadBPReg()
// ---------------
// Called att the end of every: OpcodeDecoding.cpp ExecuteDisplayList > Decode() > LoadBPReg
void BPWritten(int addr, int changes, int newval)
{
DVSTARTPROFILE();
//static int count = 0;
//ERROR_LOG("(%d) %x: %x\n", count++, addr, newval);
switch(addr)
switch (addr)
{
case BPMEM_GENMODE:
if (changes) {
@ -139,7 +135,7 @@ void BPWritten(int addr, int changes, int newval)
glDepthMask(GL_FALSE);
}
if( !bpmem.zmode.updateenable )
if (!bpmem.zmode.updateenable)
Renderer::SetRenderMode(Renderer::RM_Normal);
}
break;
@ -165,12 +161,11 @@ void BPWritten(int addr, int changes, int newval)
case BPMEM_LINEPTWIDTH:
{
float fratio = VertexShaderMngr::rawViewport[0] != 0 ? (float)Renderer::GetTargetWidth()/640.0f : 1.0f;
if( bpmem.lineptwidth.linesize > 0 ) {
glLineWidth((float)bpmem.lineptwidth.linesize*fratio/6.0f); // scale by ratio of widths
}
if( bpmem.lineptwidth.pointsize > 0 )
glPointSize((float)bpmem.lineptwidth.pointsize*fratio/6.0f);
float fratio = VertexShaderMngr::rawViewport[0] != 0 ? (float)Renderer::GetTargetWidth() / 640.0f : 1.0f;
if (bpmem.lineptwidth.linesize > 0)
glLineWidth((float)bpmem.lineptwidth.linesize * fratio / 6.0f); // scale by ratio of widths
if (bpmem.lineptwidth.pointsize > 0)
glPointSize((float)bpmem.lineptwidth.pointsize * fratio / 6.0f);
break;
}
@ -197,9 +192,9 @@ void BPWritten(int addr, int changes, int newval)
}
else glDisable(GL_BLEND);
}
if( changes & 2 ) {
if( Renderer::CanBlendLogicOp() ) {
if( bpmem.blendmode.logicopenable ) {
if (changes & 2) {
if (Renderer::CanBlendLogicOp()) {
if (bpmem.blendmode.logicopenable) {
glEnable(GL_COLOR_LOGIC_OP);
PanicAlert("Logic Op Blend : %i", bpmem.blendmode.logicmode);
glLogicOp(glLogicOpCodes[bpmem.blendmode.logicmode]);
@ -207,7 +202,7 @@ void BPWritten(int addr, int changes, int newval)
else glDisable(GL_COLOR_LOGIC_OP);
}
//else {
// if( bpmem.blendmode.logicopenable ) {
// if (bpmem.blendmode.logicopenable) {
// switch(bpmem.blendmode.logicmode) {
// case 0: // clear dst to 0
// glEnable(GL_BLEND);
@ -254,13 +249,13 @@ void BPWritten(int addr, int changes, int newval)
//if (bpmem.blendmode.dither) glEnable(GL_DITHER);
//else glDisable(GL_DITHER);
}
if( changes & 0xFE0) {
if( !bpmem.blendmode.subtract )
if (changes & 0xFE0) {
if (!bpmem.blendmode.subtract)
glBlendFunc(glSrcFactors[bpmem.blendmode.srcfactor], glDestFactors[bpmem.blendmode.dstfactor]);
}
if (changes & 0x800) {
glBlendEquation(bpmem.blendmode.subtract?GL_FUNC_REVERSE_SUBTRACT:GL_FUNC_ADD);
if( bpmem.blendmode.subtract )
if (bpmem.blendmode.subtract)
glBlendFunc(GL_ONE, GL_ONE);
else
glBlendFunc(glSrcFactors[bpmem.blendmode.srcfactor], glDestFactors[bpmem.blendmode.dstfactor]);
@ -351,8 +346,8 @@ void BPWritten(int addr, int changes, int newval)
if (changes) {
VertexManager::Flush();
((u32*)&bpmem)[addr] = newval;
if( !SetScissorRect() ) {
if( addr == BPMEM_SCISSORBR )
if (!SetScissorRect()) {
if (addr == BPMEM_SCISSORBR )
ERROR_LOG("bad scissor!\n");
}
}
@ -392,6 +387,154 @@ void BPWritten(int addr, int changes, int newval)
PixelShaderMngr::SetTevKSelChanged(addr-0xf6);
}
break;
case 0x45: //GXSetDrawDone
VertexManager::Flush();
switch (newval & 0xFF)
{
case 0x02:
g_VideoInitialize.pSetPEFinish(); // may generate interrupt
DebugLog("GXSetDrawDone SetPEFinish (value: 0x%02X)", (newval & 0xFFFF));
break;
default:
DebugLog("GXSetDrawDone ??? (value 0x%02X)", (newval & 0xFFFF));
break;
}
break;
case BPMEM_PE_TOKEN_ID:
g_VideoInitialize.pSetPEToken(static_cast<u16>(newval & 0xFFFF), FALSE);
DebugLog("SetPEToken 0x%04x", (newval & 0xFFFF));
break;
case BPMEM_PE_TOKEN_INT_ID:
g_VideoInitialize.pSetPEToken(static_cast<u16>(newval & 0xFFFF), TRUE);
DebugLog("SetPEToken + INT 0x%04x", (newval & 0xFFFF));
break;
case 0x67: // set gp metric?
break;
case 0x52:
{
DVProfileFunc _pf("LoadBPReg:swap");
VertexManager::Flush();
((u32*)&bpmem)[addr] = newval;
TRectangle rc = {
(int)(bpmem.copyTexSrcXY.x),
(int)(bpmem.copyTexSrcXY.y),
(int)((bpmem.copyTexSrcXY.x + bpmem.copyTexSrcWH.x + 1)),
(int)((bpmem.copyTexSrcXY.y + bpmem.copyTexSrcWH.y + 1))
};
//Need another rc here to get it to scale.
TRectangle multirc = {
(int)(bpmem.copyTexSrcXY.x * MValueX),
(int)(bpmem.copyTexSrcXY.y * MValueY),
(int)((bpmem.copyTexSrcXY.x * MValueX + (bpmem.copyTexSrcWH.x + 1) * MValueX)),
(int)((bpmem.copyTexSrcXY.y * MValueY + (bpmem.copyTexSrcWH.y + 1) * MValueY))
};
UPE_Copy PE_copy;
PE_copy.Hex = bpmem.triggerEFBCopy;
if (PE_copy.copy_to_xfb == 0) {
// EFB to texture
// for some reason it sets bpmem.zcontrol.pixel_format to PIXELFMT_Z24 every time a zbuffer format is given as a dest to GXSetTexCopyDst
TextureMngr::CopyRenderTargetToTexture(bpmem.copyTexDest<<5, bpmem.zcontrol.pixel_format==PIXELFMT_Z24, PE_copy.intensity_fmt>0,
(PE_copy.target_pixel_format/2)+((PE_copy.target_pixel_format&1)*8), PE_copy.half_scale>0, &rc);
}
else {
// EFB to XFB
if(g_Config.bUseXFB)
{
XFB_Write(Memory_GetPtr(bpmem.copyTexDest<<5), multirc, (bpmem.copyMipMapStrideChannels << 4), bpmem.copyTexSrcWH.y + 1, bpmem.dispcopyyscale/256.0f);
}
else
{
Renderer::Swap(multirc);
}
g_VideoInitialize.pCopiedToXFB();
}
// clearing
if (PE_copy.clear) {
// clear color
Renderer::SetRenderMode(Renderer::RM_Normal);
u32 nRestoreZBufferTarget = Renderer::GetZBufferTarget();
glViewport(0, 0, Renderer::GetTargetWidth(), Renderer::GetTargetHeight());
// Always set the scissor in case it was set by the game and has not been reset
glScissor(multirc.left, (Renderer::GetTargetHeight() - multirc.bottom),
(multirc.right - multirc.left), (multirc.bottom - multirc.top));
VertexShaderMngr::SetViewportChanged();
// since clear operations use the source rectangle, have to do regular renders (glClear clears the entire buffer)
if (bpmem.blendmode.colorupdate || bpmem.blendmode.alphaupdate || bpmem.zmode.updateenable) {
GLbitfield bits = 0;
if (bpmem.blendmode.colorupdate || bpmem.blendmode.alphaupdate) {
u32 clearColor = (bpmem.clearcolorAR<<16)|bpmem.clearcolorGB;
glClearColor(((clearColor>>16)&0xff)*(1/255.0f),((clearColor>>8)&0xff)*(1/255.0f),
((clearColor>>0)&0xff)*(1/255.0f),((clearColor>>24)&0xff)*(1/255.0f));
bits |= GL_COLOR_BUFFER_BIT;
}
if (bpmem.zmode.updateenable) {
glClearDepth((float)(bpmem.clearZValue&0xFFFFFF) / float(0xFFFFFF));
bits |= GL_DEPTH_BUFFER_BIT;
}
if (nRestoreZBufferTarget )
glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT); // don't clear ztarget here
glClear(bits);
}
if (bpmem.zmode.updateenable && nRestoreZBufferTarget) { // have to clear the target zbuffer
glDrawBuffer(GL_COLOR_ATTACHMENT1_EXT);
GL_REPORT_ERRORD();
glColorMask(GL_TRUE,GL_TRUE,GL_TRUE,GL_TRUE);
// red should probably be the LSB
glClearColor(((bpmem.clearZValue>>0)&0xff)*(1/255.0f),((bpmem.clearZValue>>8)&0xff)*(1/255.0f),
((bpmem.clearZValue>>16)&0xff)*(1/255.0f), 0);
glClear(GL_COLOR_BUFFER_BIT);
SetColorMask();
GL_REPORT_ERRORD();
}
if (nRestoreZBufferTarget) {
// restore target
GLenum s_drawbuffers[2] = {GL_COLOR_ATTACHMENT0_EXT, GL_COLOR_ATTACHMENT1_EXT};
glDrawBuffers(2, s_drawbuffers);
}
SetScissorRect(); // reset the scissor rect
}
}
break;
case 0x65: //GXLoadTlut
{
DVProfileFunc _pf("LoadBPReg:GXLoadTlut");
VertexManager::Flush();
((u32*)&bpmem)[addr] = newval;
u32 tlutTMemAddr = (newval&0x3FF)<<9;
u32 tlutXferCount = (newval&0x1FFC00)>>5;
//do the transfer!!
memcpy_gc(texMem + tlutTMemAddr, g_VideoInitialize.pGetMemoryPointer((bpmem.tlutXferSrc&0xFFFFF)<<5), tlutXferCount);
// TODO(ector) : kill all textures that use this palette
// Not sure if it's a good idea, though. For now, we hash texture palettes
}
break;
default:
switch(addr & 0xFC) //texture sampler filter
@ -490,7 +633,7 @@ void BPWritten(int addr, int changes, int newval)
break;
case 0x30:
if( changes ) {
if (changes) {
VertexManager::Flush();
((u32*)&bpmem)[addr] = newval;
PixelShaderMngr::SetTexDimsChanged((addr>>1)&0x7);
@ -591,37 +734,35 @@ bool SetScissorRect()
int xoff = bpmem.scissorOffset.x * 2 - 342;
int yoff = bpmem.scissorOffset.y * 2 - 342;
RECT rc;
int rc_left = bpmem.scissorTL.x - xoff - 342; // left = 0
rc_left *= MValueX;
if (rc_left < 0) rc_left = 0;
rc.left = bpmem.scissorTL.x - xoff - 342; // left = 0
rc.left *= MValueX;
if (rc.left < 0) rc.left = 0;
rc.top = bpmem.scissorTL.y - yoff - 342; // right = 0
rc.top *= MValueY;
if (rc.top < 0) rc.top = 0;
int rc_top = bpmem.scissorTL.y - yoff - 342; // right = 0
rc_top *= MValueY;
if (rc_top < 0) rc_top = 0;
rc.right = bpmem.scissorBR.x - xoff - 342; // right = 640
rc.right *= MValueX;
if (rc.right > 640 * MValueX) rc.right = 640 * MValueX;
int rc_right = bpmem.scissorBR.x - xoff - 342; // right = 640
rc_right *= MValueX;
if (rc_right > 640 * MValueX) rc_right = 640 * MValueX;
rc.bottom = bpmem.scissorBR.y - yoff - 342; // bottom = 480
rc.bottom *= MValueY;
if (rc.bottom > 480 * MValueY) rc.bottom = 480 * MValueY;
int rc_bottom = bpmem.scissorBR.y - yoff - 342; // bottom = 480
rc_bottom *= MValueY;
if (rc_bottom > 480 * MValueY) rc_bottom = 480 * MValueY;
/*__Log("Scissor: lt=(%d,%d), rb=(%d,%d,%i), off=(%d,%d)\n",
rc.left, rc.top,
rc.right, rc.bottom, Renderer::GetTargetHeight(),
rc_left, rc_top,
rc_right, rc_bottom, Renderer::GetTargetHeight(),
xoff, yoff
);*/
if( rc.right>=rc.left && rc.bottom>=rc.top )
if (rc_right >= rc_left && rc_bottom >= rc_top )
{
glScissor(
rc.left, // x = 0
Renderer::GetTargetHeight()-(rc.bottom), // y = 0
(rc.right-rc.left), // y = 0
(rc.bottom-rc.top) // y = 0
rc_left, // x = 0
Renderer::GetTargetHeight()-(rc_bottom), // y = 0
(rc_right-rc_left), // y = 0
(rc_bottom-rc_top) // y = 0
);
return true;
}
@ -630,183 +771,38 @@ bool SetScissorRect()
}
// =======================================================================================
// Call browser: OpcodeDecoding.cpp ExecuteDisplayList > Decode() > LoadBPReg()
// ---------------
void LoadBPReg(u32 value0)
{
DVSTARTPROFILE();
//handle the mask register
int opcode = value0 >> 24;
int oldval = ((u32*)&bpmem)[opcode];
int newval = (((u32*)&bpmem)[opcode] & ~bpmem.bpMask) | (value0 & bpmem.bpMask);
int newval = (oldval & ~bpmem.bpMask) | (value0 & bpmem.bpMask);
int changes = (oldval ^ newval) & 0xFFFFFF;
//reset the mask register
if (opcode != 0xFE)
bpmem.bpMask = 0xFFFFFF;
switch (opcode)
{
case 0x45: //GXSetDrawDone
VertexManager::Flush();
switch (value0 & 0xFF)
{
case 0x02:
g_VideoInitialize.pSetPEFinish(); // may generate interrupt
DebugLog("GXSetDrawDone SetPEFinish (value: 0x%02X)", (value0 & 0xFFFF));
break;
default:
DebugLog("GXSetDrawDone ??? (value 0x%02X)", (value0 & 0xFFFF));
break;
}
break;
case BPMEM_PE_TOKEN_ID:
g_VideoInitialize.pSetPEToken(static_cast<u16>(value0 & 0xFFFF), FALSE);
DebugLog("SetPEToken 0x%04x", (value0 & 0xFFFF));
break;
case BPMEM_PE_TOKEN_INT_ID:
g_VideoInitialize.pSetPEToken(static_cast<u16>(value0 & 0xFFFF), TRUE);
DebugLog("SetPEToken + INT 0x%04x", (value0 & 0xFFFF));
break;
case 0x67: // set gp metric?
break;
case 0x52:
{
DVProfileFunc _pf("LoadBPReg:swap");
VertexManager::Flush();
((u32*)&bpmem)[opcode] = newval;
TRectangle rc = {
(int)(bpmem.copyTexSrcXY.x),
(int)(bpmem.copyTexSrcXY.y),
(int)((bpmem.copyTexSrcXY.x + bpmem.copyTexSrcWH.x + 1)),
(int)((bpmem.copyTexSrcXY.y + bpmem.copyTexSrcWH.y + 1))
};
//Need another rc here to get it to scale.
TRectangle multirc = {
(int)(bpmem.copyTexSrcXY.x * MValueX),
(int)(bpmem.copyTexSrcXY.y * MValueY),
(int)((bpmem.copyTexSrcXY.x * MValueX + (bpmem.copyTexSrcWH.x + 1) * MValueX)),
(int)((bpmem.copyTexSrcXY.y * MValueY + (bpmem.copyTexSrcWH.y + 1) * MValueY))
};
UPE_Copy PE_copy;
PE_copy.Hex = bpmem.triggerEFBCopy;
if (PE_copy.copy_to_xfb == 0) {
// EFB to texture
// for some reason it sets bpmem.zcontrol.pixel_format to PIXELFMT_Z24 every time a zbuffer format is given as a dest to GXSetTexCopyDst
TextureMngr::CopyRenderTargetToTexture(bpmem.copyTexDest<<5, bpmem.zcontrol.pixel_format==PIXELFMT_Z24, PE_copy.intensity_fmt>0,
(PE_copy.target_pixel_format/2)+((PE_copy.target_pixel_format&1)*8), PE_copy.half_scale>0, &rc);
}
else {
// EFB to XFB
if(g_Config.bUseXFB)
{
XFB_Write(Memory_GetPtr(bpmem.copyTexDest<<5), multirc, (bpmem.copyMipMapStrideChannels << 4), bpmem.copyTexSrcWH.y + 1, bpmem.dispcopyyscale/256.0f);
}
else
{
Renderer::Swap(multirc);
}
g_VideoInitialize.pCopiedToXFB();
}
// clearing
if (PE_copy.clear) {
// clear color
Renderer::SetRenderMode(Renderer::RM_Normal);
u32 nRestoreZBufferTarget = Renderer::GetZBufferTarget();
glViewport(0, 0, Renderer::GetTargetWidth(), Renderer::GetTargetHeight());
// Always set the scissor in case it was set by the game and has not been reset
glScissor(multirc.left, (Renderer::GetTargetHeight() - multirc.bottom),
(multirc.right - multirc.left), (multirc.bottom - multirc.top));
VertexShaderMngr::SetViewportChanged();
// since clear operations use the source rectangle, have to do regular renders (glClear clears the entire buffer)
if( bpmem.blendmode.colorupdate || bpmem.blendmode.alphaupdate || bpmem.zmode.updateenable ) {
GLbitfield bits = 0;
if( bpmem.blendmode.colorupdate || bpmem.blendmode.alphaupdate ) {
u32 clearColor = (bpmem.clearcolorAR<<16)|bpmem.clearcolorGB;
glClearColor(((clearColor>>16)&0xff)*(1/255.0f),((clearColor>>8)&0xff)*(1/255.0f),
((clearColor>>0)&0xff)*(1/255.0f),((clearColor>>24)&0xff)*(1/255.0f));
bits |= GL_COLOR_BUFFER_BIT;
}
if( bpmem.zmode.updateenable ) {
glClearDepth((float)(bpmem.clearZValue&0xFFFFFF) / float(0xFFFFFF));
bits |= GL_DEPTH_BUFFER_BIT;
}
if( nRestoreZBufferTarget )
glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT); // don't clear ztarget here
glClear(bits);
}
if (bpmem.zmode.updateenable && nRestoreZBufferTarget ) { // have to clear the target zbuffer
glDrawBuffer(GL_COLOR_ATTACHMENT1_EXT);
GL_REPORT_ERRORD();
glColorMask(GL_TRUE,GL_TRUE,GL_TRUE,GL_TRUE);
// red should probably be the LSB
glClearColor(((bpmem.clearZValue>>0)&0xff)*(1/255.0f),((bpmem.clearZValue>>8)&0xff)*(1/255.0f),
((bpmem.clearZValue>>16)&0xff)*(1/255.0f), 0);
glClear(GL_COLOR_BUFFER_BIT);
SetColorMask();
GL_REPORT_ERRORD();
}
if( nRestoreZBufferTarget ) {
// restore target
GLenum s_drawbuffers[2] = {GL_COLOR_ATTACHMENT0_EXT, GL_COLOR_ATTACHMENT1_EXT};
glDrawBuffers(2, s_drawbuffers);
}
SetScissorRect(); // reset the scissor rect
}
}
break;
case 0x65: //GXLoadTlut
{
DVProfileFunc _pf("LoadBPReg:GXLoadTlut");
VertexManager::Flush();
((u32*)&bpmem)[opcode] = newval;
u32 tlutTMemAddr = (value0&0x3FF)<<9;
u32 tlutXferCount = (value0&0x1FFC00)>>5;
//do the transfer!!
memcpy_gc(texMem + tlutTMemAddr, g_VideoInitialize.pGetMemoryPointer((bpmem.tlutXferSrc&0xFFFFF)<<5), tlutXferCount);
// TODO(ector) : kill all textures that use this palette
// Not sure if it's a good idea, though. For now, we hash texture palettes
}
break;
}
//notify the video handling so it can update render states
BPWritten(opcode, changes, newval);
//((u32*)&bpmem)[opcode] = newval;
}
// =======================================================================================
// Never called?
// ---------------
// Never called? Should probably be called when loading a saved state.
// Needs testing though.
void BPReload()
{
for (int i=0; i<254; i++)
BPWritten(i, 0xFFFFFF, ((u32*)&bpmem)[i]);
for (int i = 0; i < 254; i++)
{
switch (i) {
case 0x65:
case 0x45: //GXSetDrawDone
case BPMEM_PE_TOKEN_ID:
case BPMEM_PE_TOKEN_INT_ID:
case 0x67: // set gp metric?
case 0x52:
break;
default:
BPWritten(i, 0xFFFFFF, ((u32*)&bpmem)[i]);
}
}
}

1
Source/Plugins/Plugin_VideoOGL/Src/BPStructs.h
View File

@ -21,7 +21,6 @@
#include "BPMemory.h"
void BPInit();
//bool BPWritten(int addr, int changes);
void LoadBPReg(u32 value0);
void SetColorMask();

14
Source/Plugins/Plugin_VideoOGL/Src/Debugger/Debugger.cpp
View File

@ -1,23 +1,19 @@
//////////////////////////////////////////////////////////////////////////////////////////
//
// Licensetype: GNU General Public License (GPL)
//
// Copyright (C) 2003-2008 Dolphin Project.
// 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, version 2.0.
//
// 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 2.0 for more details.
//
// A copy of the GPL 2.0 should have been included with the program.
// If not, see http://www.gnu.org/licenses/
//
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
//
//////////////////////////////////////////////////////////////////////////////////////////
#include "../Globals.h"
#include "Debugger.h"

14
Source/Plugins/Plugin_VideoOGL/Src/Debugger/Debugger.h
View File

@ -1,23 +1,19 @@
//////////////////////////////////////////////////////////////////////////////////////////
//
// Licensetype: GNU General Public License (GPL)
//
// Copyright (C) 2003-2008 Dolphin Project.
// 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, version 2.0.
//
// 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 2.0 for more details.
//
// A copy of the GPL 2.0 should have been included with the program.
// If not, see http://www.gnu.org/licenses/
//
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
//
//////////////////////////////////////////////////////////////////////////////////////////
#ifndef __CDebugger_h__
#define __CDebugger_h__

32
Source/Plugins/Plugin_VideoOGL/Src/Debugger/PBView.cpp
View File

@ -15,10 +15,6 @@
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
// ---------------------------------------------------------------------------------------
// includes
// -----------------
#include "../Globals.h"
#include "PBView.h"
@ -27,24 +23,11 @@
#include <stdio.h>
#include <stdlib.h>
// ---------------------------------------------------------------------------------------
// external declarations
// -----------------
extern const char* GetGRPName(unsigned int index);
// ---------------------------------------------------------------------------------------
// No buttons or events so far
// -----------------
BEGIN_EVENT_TABLE(CPBView, wxListCtrl)
END_EVENT_TABLE()
// =======================================================================================
// The main wxListCtrl
// -------------
CPBView::CPBView(wxWindow* parent, const wxWindowID id, const wxPoint& pos, const wxSize& size, long style)
: wxListCtrl(parent, id, pos, size, style)
{
@ -54,7 +37,6 @@ CPBView::CPBView(wxWindow* parent, const wxWindowID id, const wxPoint& pos, cons
for (int i = 0; i < 1; i++)
{
// Print values from 0 to 63
char buffer [33];
sprintf(buffer, "%02i", i);
@ -72,18 +54,14 @@ CPBView::CPBView(wxWindow* parent, const wxWindowID id, const wxPoint& pos, cons
Refresh();
}
void
CPBView::Update()
void CPBView::Update()
{
Refresh();
}
bool
CPBView::MSWDrawSubItem(wxPaintDC& rPainDC, int item, int subitem)
bool CPBView::MSWDrawSubItem(wxPaintDC& rPainDC, int item, int subitem)
{
bool Result = false;
@ -135,13 +113,11 @@ CPBView::MSWDrawSubItem(wxPaintDC& rPainDC, int item, int subitem)
rPainDC.DrawText(text, 10, 4);
#endif
return(true);
return true;
}
else
{
// what does this mean?
return(Result);
return Result;
}
}

26
Source/Plugins/Plugin_VideoOGL/Src/Debugger/PBView.h
View File

@ -23,25 +23,17 @@
#include "Common.h"
class CPBView
: public wxListCtrl
class CPBView : public wxListCtrl
{
public:
public:
CPBView(wxWindow* parent, const wxWindowID id, const wxPoint& pos, const wxSize& size, long style);
void Update();
u32 m_CachedRegs[10][10];
CPBView(wxWindow* parent, const wxWindowID id, const wxPoint& pos, const wxSize& size, long style);
void Update();
u32 m_CachedRegs[10][10];
private:
DECLARE_EVENT_TABLE()
bool m_CachedRegHasChanged[64];
virtual bool MSWDrawSubItem(wxPaintDC& rPainDC, int item, int subitem);
private:
DECLARE_EVENT_TABLE()
bool m_CachedRegHasChanged[64];
virtual bool MSWDrawSubItem(wxPaintDC& rPainDC, int item, int subitem);
};
#endif

21
Source/Plugins/Plugin_VideoOGL/Src/GLInit.cpp
View File

@ -19,9 +19,18 @@
#include "IniFile.h"
#include "svnrev.h"
#include "Render.h"
#if defined(_WIN32)
#include "OS/Win32.h"
#else
struct RECT
{
int left, top;
int right, bottom;
};
#endif
#include "GLInit.h"
#ifndef USE_SDL
@ -33,16 +42,12 @@
// Handles OpenGL and the window
// ---------------------------------------------------------------------------------------
// externals
// -------------
int gleft, gright, gtop, gbottom;
int nBackbufferWidth, nBackbufferHeight; // screen width
int nXoff, nYoff; // screen offset
float AR; // aspect ratio
#ifndef _WIN32
GLWindow GLWin;
#endif
@ -104,7 +109,6 @@ BOOL Callback_PeekMessages()
#endif
}
void UpdateFPSDisplay(const char *text)
{
char temp[512];
@ -116,10 +120,8 @@ void UpdateFPSDisplay(const char *text)
}
// =======================================================================================
// Create window. Called from main.cpp
// ----------------
bool OpenGL_Create(SVideoInitialize &_VideoInitialize, int _iwidth, int _iheight)
{
int _twidth, _theight;
@ -186,7 +188,6 @@ bool OpenGL_Create(SVideoInitialize &_VideoInitialize, int _iwidth, int _iheight
//sprintf(buff, "%i %i %d %d %d", nBackbufferWidth, nBackbufferHeight, Max, MValueX, MValueY);
//MessageBox(0, buff, "", 0);
#if USE_SDL
//init sdl video
if (SDL_Init(SDL_INIT_VIDEO) < 0) {
@ -236,7 +237,7 @@ bool OpenGL_Create(SVideoInitialize &_VideoInitialize, int _iwidth, int _iheight
dmScreenSettings.dmFields = DM_BITSPERPEL|DM_PELSWIDTH|DM_PELSHEIGHT;
// Try To Set Selected Mode And Get Results. NOTE: CDS_FULLSCREEN Gets Rid Of Start Bar.
if (ChangeDisplaySettings(&dmScreenSettings,CDS_FULLSCREEN)!=DISP_CHANGE_SUCCESSFUL)
if (ChangeDisplaySettings(&dmScreenSettings, CDS_FULLSCREEN) != DISP_CHANGE_SUCCESSFUL)
{
if (MessageBox(NULL,"The Requested Fullscreen Mode Is Not Supported By\nYour Video Card. Use Windowed Mode Instead?","NeHe GL",MB_YESNO|MB_ICONEXCLAMATION)==IDYES)
g_Config.bFullscreen = false;
@ -619,7 +620,7 @@ void OpenGL_Update()
float Max = (FactorW < FactorH) ? FactorH : FactorW;
AR = (float)nBackbufferWidth / (float)nBackbufferHeight;
if(g_Config.bStretchToFit && g_Config.renderToMainframe)
if (g_Config.bStretchToFit && g_Config.renderToMainframe)
{
MValueX = 1;
MValueY = 1;

118
Source/Plugins/Plugin_VideoOGL/Src/Globals.cpp
View File

@ -15,7 +15,6 @@
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
#include <wx/wx.h>
#include <wx/filepicker.h>
#include <wx/notebook.h>
@ -30,8 +29,6 @@
#include "IniFile.h"
#include <assert.h>
float MValueX, MValueY; // Since it can Stretch to fit Window, we need two different multiplication values//
int frameCount;
Config g_Config;
Statistics stats;
@ -52,10 +49,6 @@ void Config::Load()
IniFile iniFile;
iniFile.Load("gfx_opengl.ini");
iniFile.Get("Hardware", "Adapter", &iAdapter, 0);
if (iAdapter == -1)
iAdapter = 0;
// get resolution
iniFile.Get("Hardware", "WindowedRes", &temp, 0);
if(temp.empty())
@ -71,7 +64,6 @@ void Config::Load()
iniFile.Get("Settings", "ShowFPS", &bShowFPS, false); // Settings
iniFile.Get("Settings", "OverlayStats", &bOverlayStats, false);
iniFile.Get("Settings", "Postprocess", &iPostprocessEffect, 0);
iniFile.Get("Settings", "DLOptimize", &iCompileDLsLevel, 0);
iniFile.Get("Settings", "DumpTextures", &bDumpTextures, 0);
iniFile.Get("Settings", "ShowShaderErrors", &bShowShaderErrors, 0);
@ -81,7 +73,7 @@ void Config::Load()
iMultisampleMode = 1;
std::string s;
iniFile.Get("Settings", "TexDumpPath", &s, 0);
if( s.size() < sizeof(texDumpPath) )
if (s.size() < sizeof(texDumpPath) )
strcpy(texDumpPath, s.c_str());
else {
strncpy(texDumpPath, s.c_str(), sizeof(texDumpPath)-1);
@ -102,7 +94,6 @@ void Config::Save()
{
IniFile iniFile;
iniFile.Load("gfx_opengl.ini");
iniFile.Set("Hardware", "Adapter", iAdapter);
iniFile.Set("Hardware", "WindowedRes", iWindowedRes);
iniFile.Set("Hardware", "FullscreenRes", iFSResolution);
iniFile.Set("Hardware", "Fullscreen", bFullscreen);
@ -110,7 +101,6 @@ void Config::Save()
iniFile.Set("Settings", "ShowFPS", bShowFPS);
iniFile.Set("Settings", "OverlayStats", bOverlayStats);
iniFile.Set("Settings", "Postprocess", iPostprocessEffect);
iniFile.Set("Settings", "DLOptimize", iCompileDLsLevel);
iniFile.Set("Settings", "DumpTextures", bDumpTextures);
iniFile.Set("Settings", "ShowShaderErrors", bShowShaderErrors);
@ -129,93 +119,19 @@ void Config::Save()
iniFile.Save("gfx_opengl.ini");
}
#if defined(_MSC_VER)
#pragma pack(push, 1)
#endif
struct TGA_HEADER
{
u8 identsize; // size of ID field that follows 18 u8 header (0 usually)
u8 colourmaptype; // type of colour map 0=none, 1=has palette
u8 imagetype; // type of image 0=none,1=indexed,2=rgb,3=grey,+8=rle packed
s16 colourmapstart; // first colour map entry in palette
s16 colourmaplength; // number of colours in palette
u8 colourmapbits; // number of bits per palette entry 15,16,24,32
s16 xstart; // image x origin
s16 ystart; // image y origin
s16 width; // image width in pixels
s16 height; // image height in pixels
u8 bits; // image bits per pixel 8,16,24,32
u8 descriptor; // image descriptor bits (vh flip bits)
// pixel data follows header
};
#if defined(_MSC_VER)
#pragma pack(pop)
#endif
bool SaveTGA(const char* filename, int width, int height, void* pdata)
{
TGA_HEADER hdr;
FILE* f = fopen(filename, "wb");
if (f == NULL)
return false;
_assert_( sizeof(TGA_HEADER) == 18 && sizeof(hdr) == 18 );
memset(&hdr, 0, sizeof(hdr));
hdr.imagetype = 2;
hdr.bits = 32;
hdr.width = width;
hdr.height = height;
hdr.descriptor |= 8|(1<<5); // 8bit alpha, flip vertical
fwrite(&hdr, sizeof(hdr), 1, f);
fwrite(pdata, width*height*4, 1, f);
fclose(f);
return true;
}
bool SaveTexture(const char* filename, u32 textarget, u32 tex, int width, int height)
{
GL_REPORT_ERRORD();
std::vector<u32> data(width*height);
glBindTexture(textarget, tex);
glGetTexImage(textarget, 0, GL_BGRA, GL_UNSIGNED_BYTE, &data[0]);
GLenum err;
GL_REPORT_ERROR();
if (err != GL_NO_ERROR) {
return false;
}
return SaveTGA(filename, width, height, &data[0]);
}
bool SaveData(const char* filename, const char* data)
{
FILE* f = fopen(filename, "wb");
if (f == NULL)
return false;
fwrite(data, strlen(data), 1, f);
fclose(f);
return true;
}
#ifdef _WIN32
// The one for Linux is in Linux/Linux.cpp
static HANDLE hConsole = NULL;
void OpenConsole() {
void OpenConsole()
{
COORD csize;
CONSOLE_SCREEN_BUFFER_INFO csbiInfo;
SMALL_RECT srect;
if (hConsole) return;
if (hConsole)
return;
AllocConsole();
SetConsoleTitle("Opengl Plugin Output");
@ -234,18 +150,18 @@ void OpenConsole() {
hConsole = GetStdHandle(STD_OUTPUT_HANDLE);
}
void CloseConsole() {
if (hConsole == NULL) return;
FreeConsole(); hConsole = NULL;
void CloseConsole()
{
if (hConsole == NULL)
return;
FreeConsole();
hConsole = NULL;
}
#endif
static FILE* pfLog = NULL;
void __Log(const char *fmt, ...)
{
char* Msg = (char*)alloca(strlen(fmt)+512);
va_list ap;
@ -255,9 +171,10 @@ void __Log(const char *fmt, ...)
g_VideoInitialize.pLog(Msg, FALSE);
if( pfLog == NULL ) pfLog = fopen("Logs/oglgfx.txt", "w");
if (pfLog == NULL)
pfLog = fopen("Logs/oglgfx.txt", "w");
if( pfLog != NULL )
if (pfLog != NULL)
fwrite(Msg, strlen(Msg), 1, pfLog);
#ifdef _WIN32
DWORD tmp;
@ -265,12 +182,10 @@ void __Log(const char *fmt, ...)
#else
//printf("%s", Msg);
#endif
}
void __Log(int type, const char *fmt, ...)
{
char* Msg = (char*)alloca(strlen(fmt)+512);
va_list ap;
@ -284,5 +199,4 @@ void __Log(int type, const char *fmt, ...)
DWORD tmp;
WriteConsole(hConsole, Msg, (DWORD)strlen(Msg), &tmp, 0);
#endif
}

34
Source/Plugins/Plugin_VideoOGL/Src/Globals.h
View File

@ -67,17 +67,8 @@ inline unsigned long timeGetTime()
return (unsigned long)(t.time*1000+t.millitm);
}
struct RECT
{
int left, top;
int right, bottom;
};
#endif // linux basic definitions
#include <Cg/cg.h>
#include <Cg/cgGL.h>
#ifndef GL_DEPTH24_STENCIL8_EXT // allows FBOs to support stencils
#define GL_DEPTH_STENCIL_EXT 0x84F9
#define GL_UNSIGNED_INT_24_8_EXT 0x84FA
@ -85,8 +76,6 @@ struct RECT
#define GL_TEXTURE_STENCIL_SIZE_EXT 0x88F1
#endif
extern float MValueX, MValueY;
#define ERROR_LOG __Log
#if defined(_DEBUG) || defined(DEBUGFAST)
@ -107,11 +96,6 @@ extern float MValueX, MValueY;
#define GL_REPORT_ERRORD()
#endif
#ifndef ARRAY_SIZE
#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
#endif
extern int frameCount;
#define CONF_LOG 1
#define CONF_PRIMLOG 2
@ -145,19 +129,11 @@ struct Config
bool bDumpTextures;
char texDumpPath[280];
//currently unused:
int iLog; // CONF_ bits
int iSaveTargetId;
int iAdapter;
char psProfile[16];
char vsProfile[16];
int iPostprocessEffect;
//currently unused:
int iCompileDLsLevel;
bool bPreUpscale;
int iPreUpscaleFilter;
bool bTruform;
int iTruformLevel;
bool bOldCard;
bool bWireFrame;
bool bShowShaderErrors;
};
@ -199,7 +175,6 @@ struct Statistics
int numDLPrims;
int numPrims;
int numShaderChanges;
int numBadCommands; //hope this always is zero ;)
int numDListsCalled;
};
@ -226,11 +201,6 @@ void __Log(const char *format, ...);
void __Log(int type, const char *format, ...);
void HandleGLError();
void InitLUTs();
bool SaveTGA(const char* filename, int width, int height, void* pdata);
bool SaveTexture(const char* filename, u32 textarget, u32 tex, int width, int height);
bool SaveData(const char* filename, const char* pdata);
#if defined(_MSC_VER) && !defined(__x86_64__) && !defined(_M_X64)
void * memcpy_amd(void *dest, const void *src, size_t n);
unsigned char memcmp_mmx(const void* src1, const void* src2, int cmpsize);

97
Source/Plugins/Plugin_VideoOGL/Src/ImageWrite.cpp Normal file
View File

@ -0,0 +1,97 @@
// Copyright (C) 2003-2008 Dolphin Project.
// 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, version 2.0.
// 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 2.0 for more details.
// A copy of the GPL 2.0 should have been included with the program.
// If not, see http://www.gnu.org/licenses/
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
#include "Globals.h"
#include <stdio.h>
#include "ImageWrite.h"
#if defined(_MSC_VER)
#pragma pack(push, 1)
#endif
struct TGA_HEADER
{
u8 identsize; // size of ID field that follows 18 u8 header (0 usually)
u8 colourmaptype; // type of colour map 0=none, 1=has palette
u8 imagetype; // type of image 0=none,1=indexed,2=rgb,3=grey,+8=rle packed
s16 colourmapstart; // first colour map entry in palette
s16 colourmaplength; // number of colours in palette
u8 colourmapbits; // number of bits per palette entry 15,16,24,32
s16 xstart; // image x origin
s16 ystart; // image y origin
s16 width; // image width in pixels
s16 height; // image height in pixels
u8 bits; // image bits per pixel 8,16,24,32
u8 descriptor; // image descriptor bits (vh flip bits)
// pixel data follows header
};
#if defined(_MSC_VER)
#pragma pack(pop)
#endif
bool SaveTGA(const char* filename, int width, int height, void* pdata)
{
TGA_HEADER hdr;
FILE* f = fopen(filename, "wb");
if (f == NULL)
return false;
_assert_(sizeof(TGA_HEADER) == 18 && sizeof(hdr) == 18);
memset(&hdr, 0, sizeof(hdr));
hdr.imagetype = 2;
hdr.bits = 32;
hdr.width = width;
hdr.height = height;
hdr.descriptor |= 8|(1<<5); // 8bit alpha, flip vertical
fwrite(&hdr, sizeof(hdr), 1, f);
fwrite(pdata, width * height * 4, 1, f);
fclose(f);
return true;
}
bool SaveTexture(const char* filename, u32 textarget, u32 tex, int width, int height)
{
GL_REPORT_ERRORD();
std::vector<u32> data(width * height);
glBindTexture(textarget, tex);
glGetTexImage(textarget, 0, GL_BGRA, GL_UNSIGNED_BYTE, &data[0]);
GLenum err;
GL_REPORT_ERROR();
if (err != GL_NO_ERROR)
{
return false;
}
return SaveTGA(filename, width, height, &data[0]);
}
bool SaveData(const char* filename, const char* data)
{
FILE *f = fopen(filename, "wb");
if (!f)
return false;
fwrite(data, strlen(data), 1, f);
fclose(f);
return true;
}

25
Source/Plugins/Plugin_VideoOGL/Src/ImageWrite.h Normal file
View File

@ -0,0 +1,25 @@
// Copyright (C) 2003-2008 Dolphin Project.
// 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, version 2.0.
// 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 2.0 for more details.
// A copy of the GPL 2.0 should have been included with the program.
// If not, see http://www.gnu.org/licenses/
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
#ifndef _IMAGEWRITE_H
#define _IMAGEWRITE_H
bool SaveTGA(const char* filename, int width, int height, void* pdata);
bool SaveTexture(const char* filename, u32 textarget, u32 tex, int width, int height);
bool SaveData(const char* filename, const char* pdata);
#endif // _IMAGEWRITE_H

4
Source/Plugins/Plugin_VideoOGL/Src/OpcodeDecoding.cpp
View File

@ -28,8 +28,8 @@
#include "Profiler.h"
#include "OpcodeDecoding.h"
#include "VertexLoader.h"
#include "VertexManager.h"
#include "VertexShaderManager.h"
#include "TextureMngr.h"
#include "BPStructs.h"
#include "Fifo.h"
@ -84,7 +84,7 @@ void ExecuteDisplayList(u32 address, u32 size)
bool FifoCommandRunnable(void)
{
u32 iBufferSize = (u32)(FAKE_GetFifoEndPtr()-g_pVideoData);
u32 iBufferSize = (u32)(FAKE_GetFifoEndPtr() - g_pVideoData);
if (iBufferSize == 0)
return false; // can't peek

252
Source/Plugins/Plugin_VideoOGL/Src/PixelShader.cpp
View File

@ -40,7 +40,7 @@ void WrapNonPow2Tex(char* &p, const char* var, int texmap, u32 texture_mask);
void WriteAlphaCompare(char *&p, int num, int comp);
bool WriteAlphaTest(char *&p);
const float epsilon = 1.0f/255.0f;
const float epsilon8bit = 1.0f / 255.0f;
static const char *tevKSelTableC[] = // KCSEL
{
@ -281,8 +281,8 @@ char *GeneratePixelShader(u32 texture_mask, bool has_zbuffer_target, bool bRende
int numTexgen = bpmem.genMode.numtexgens;
char *p = text;
WRITE(p,"//Pixel Shader for TEV stages\n");
WRITE(p,"//%i TEV stages, %i texgens, %i IND stages\n",
WRITE(p, "//Pixel Shader for TEV stages\n");
WRITE(p, "//%i TEV stages, %i texgens, %i IND stages\n",
numStages,numTexgen,bpmem.genMode.numindstages);
bool bRenderZ = has_zbuffer_target && bpmem.zmode.updateenable;
@ -293,24 +293,24 @@ char *GeneratePixelShader(u32 texture_mask, bool has_zbuffer_target, bool bRende
assert( !bRenderZToCol0 || bRenderZ );
int ztexcoord = -1;
if( bInputZ )
if (bInputZ)
ztexcoord = numTexgen == 0 ? 0 : numTexgen-1;
int nIndirectStagesUsed = 0;
if( bpmem.genMode.numindstages > 0 ) {
for(int i = 0; i < numStages; ++i) {
if( bpmem.tevind[i].IsActive() && bpmem.tevind[i].bt < bpmem.genMode.numindstages ) {
if (bpmem.genMode.numindstages > 0) {
for (int i = 0; i < numStages; ++i) {
if (bpmem.tevind[i].IsActive() && bpmem.tevind[i].bt < bpmem.genMode.numindstages) {
nIndirectStagesUsed |= 1<<bpmem.tevind[i].bt;
}
}
}
// samplers
if( texture_mask ) {
WRITE(p,"uniform samplerRECT ");
if (texture_mask) {
WRITE(p, "uniform samplerRECT ");
bool bfirst = true;
for(int i = 0; i < 8; ++i) {
if( texture_mask & (1<<i) ) {
for (int i = 0; i < 8; ++i) {
if (texture_mask & (1<<i)) {
WRITE(p, "%s samp%d : register(s%d)", bfirst?"":",", i, i);
bfirst = false;
}
@ -318,11 +318,11 @@ char *GeneratePixelShader(u32 texture_mask, bool has_zbuffer_target, bool bRende
WRITE(p, ";\n");
}
if( texture_mask != 0xff ) {
WRITE(p,"uniform sampler2D ");
if (texture_mask != 0xff) {
WRITE(p, "uniform sampler2D ");
bool bfirst = true;
for(int i = 0; i < 8; ++i) {
if( !(texture_mask & (1<<i)) ) {
for (int i = 0; i < 8; ++i) {
if (!(texture_mask & (1<<i))) {
WRITE(p, "%s samp%d : register(s%d)", bfirst?"":",",i, i);
bfirst = false;
}
@ -332,7 +332,7 @@ char *GeneratePixelShader(u32 texture_mask, bool has_zbuffer_target, bool bRende
WRITE(p, "\n");
WRITE(p,"uniform float4 "I_COLORS"[4] : register(c%d);\n", C_COLORS);
WRITE(p, "uniform float4 "I_COLORS"[4] : register(c%d);\n", C_COLORS);
WRITE(p, "uniform float4 "I_KCOLORS"[4] : register(c%d);\n", C_KCOLORS);
WRITE(p, "uniform float4 "I_ALPHA"[1] : register(c%d);\n", C_ALPHA);
WRITE(p, "uniform float4 "I_TEXDIMS"[8] : register(c%d);\n", C_TEXDIMS);
@ -340,65 +340,65 @@ char *GeneratePixelShader(u32 texture_mask, bool has_zbuffer_target, bool bRende
WRITE(p, "uniform float4 "I_INDTEXSCALE"[2] : register(c%d);\n", C_INDTEXSCALE);
WRITE(p, "uniform float4 "I_INDTEXMTX"[6] : register(c%d);\n", C_INDTEXMTX);
WRITE(p,"void main(\n");
WRITE(p, "void main(\n");
WRITE(p, "out half4 ocol0 : COLOR0,\n");
if( bRenderZ && !bRenderZToCol0 )
if (bRenderZ && !bRenderZToCol0 )
WRITE(p, "out half4 ocol1 : COLOR1,\n");
if( bOutputZ )
WRITE(p," out float depth : DEPTH,\n");
if (bOutputZ )
WRITE(p, " out float depth : DEPTH,\n");
// if zcoord might come from vertex shader in texcoord
if( bInputZ ) {
if (bInputZ) {
if (numTexgen) {
for(int i = 0; i < numTexgen; ++i)
WRITE(p," in float%d uv%d : TEXCOORD%d, \n", i==ztexcoord?4:3, i,i);
for (int i = 0; i < numTexgen; ++i)
WRITE(p, " in float%d uv%d : TEXCOORD%d, \n", i==ztexcoord?4:3, i,i);
}
else
WRITE(p," in float4 uv0 : TEXCOORD0,"); //HACK
WRITE(p, " in float4 uv0 : TEXCOORD0,"); //HACK
}
else {
if (numTexgen) {
for(int i = 0; i < numTexgen; ++i)
WRITE(p," in float3 uv%d : TEXCOORD%d,\n",i,i);
for (int i = 0; i < numTexgen; ++i)
WRITE(p, " in float3 uv%d : TEXCOORD%d,\n",i,i);
}
else
WRITE(p," in float3 uv0 : TEXCOORD0,\n"); //HACK
WRITE(p, " in float3 uv0 : TEXCOORD0,\n"); //HACK
}
WRITE(p, " in float4 colors[2] : COLOR0){\n");
char* pmainstart = p;
WRITE(p,"float4 c0="I_COLORS"[1],c1="I_COLORS"[2],c2="I_COLORS"[3],prev=float4(0.0f,0.0f,0.0f,0.0f),textemp,rastemp,konsttemp=float4(0.0f,0.0f,0.0f,0.0f);\n"
WRITE(p, "float4 c0="I_COLORS"[1],c1="I_COLORS"[2],c2="I_COLORS"[3],prev=float4(0.0f,0.0f,0.0f,0.0f),textemp,rastemp,konsttemp=float4(0.0f,0.0f,0.0f,0.0f);\n"
"float3 comp16 = float3(1,255,0), comp24 = float3(1,255,255*255);\n"
"float4 alphabump=0;\n"
"float3 tevcoord;\n"
"float2 wrappedcoord, tempcoord;\n");
//if( bOutputZ ) WRITE(p," float depth;\n");
//if (bOutputZ ) WRITE(p, " float depth;\n");
// WRITE(p, "return 1;}\n");
// return PixelShaderMngr::CompilePixelShader(ps, text);
// indirect texture map lookup
for(u32 i = 0; i < bpmem.genMode.numindstages; ++i) {
if( nIndirectStagesUsed & (1<<i) ) {
if (nIndirectStagesUsed & (1<<i)) {
// perform indirect texture map lookup
// note that we have to scale by the regular texture map's coordinates since this is a texRECT call
// (and we have to match with the game's texscale calls)
int texcoord = bpmem.tevindref.getTexCoord(i);
if( texture_mask & (1<<bpmem.tevindref.getTexMap(i)) ) {
if (texture_mask & (1<<bpmem.tevindref.getTexMap(i))) {
// TODO: I removed a superfluous argument, please check that the resulting expression is correct. (mthuurne 2008-08-27)
WRITE(p, "float2 induv%d=uv%d.xy * "I_INDTEXSCALE"[%d].%s;\n", i, texcoord, i/2, (i&1)?"zw":"xy"); //, bpmem.tevindref.getTexMap(i)
char str[16];
sprintf(str, "induv%d", i);
WrapNonPow2Tex(p, str, bpmem.tevindref.getTexMap(i), texture_mask);
WRITE(p,"float3 indtex%d=texRECT(samp%d,induv%d.xy).abg;\n", i, bpmem.tevindref.getTexMap(i), i);
WRITE(p, "float3 indtex%d=texRECT(samp%d,induv%d.xy).abg;\n", i, bpmem.tevindref.getTexMap(i), i);
}
else {
WRITE(p,"float3 indtex%d=tex2D(samp%d,uv%d.xy*"I_INDTEXSCALE"[%d].%s).abg;\n", i, bpmem.tevindref.getTexMap(i), texcoord, i/2, (i&1)?"zw":"xy");
WRITE(p, "float3 indtex%d=tex2D(samp%d,uv%d.xy*"I_INDTEXSCALE"[%d].%s).abg;\n", i, bpmem.tevindref.getTexMap(i), texcoord, i/2, (i&1)?"zw":"xy");
}
}
}
@ -406,9 +406,9 @@ char *GeneratePixelShader(u32 texture_mask, bool has_zbuffer_target, bool bRende
for (int i = 0; i < numStages; i++)
WriteStage(p, i, texture_mask); //build the equation for this stage
if( bOutputZ ) {
if (bOutputZ) {
// use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format...
if( bpmem.ztex2.op == ZTEXTURE_ADD ) {
if (bpmem.ztex2.op == ZTEXTURE_ADD) {
WRITE(p, "depth = frac(dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w + uv%d.w);\n", ztexcoord);
}
else {
@ -417,40 +417,40 @@ char *GeneratePixelShader(u32 texture_mask, bool has_zbuffer_target, bool bRende
}
}
//if( bpmem.genMode.numindstages ) WRITE(p, "prev.rg = indtex0.xy;\nprev.b = 0;\n");
//if (bpmem.genMode.numindstages ) WRITE(p, "prev.rg = indtex0.xy;\nprev.b = 0;\n");
if( !WriteAlphaTest(p) ) {
if (!WriteAlphaTest(p)) {
// alpha test will always fail, so restart the shader and just make it an empty function
p = pmainstart;
WRITE(p, "discard;\n");
WRITE(p, "ocol0 = 0;\n");
}
else {
if( !bRenderZToCol0 ) {
if (!bRenderZToCol0) {
if (bpmem.dstalpha.enable)
WRITE(p," ocol0 = float4(prev.rgb,"I_ALPHA"[0].w);\n");
WRITE(p, " ocol0 = float4(prev.rgb,"I_ALPHA"[0].w);\n");
else
WRITE(p," ocol0 = prev;\n");
WRITE(p, " ocol0 = prev;\n");
}
}
if( bRenderZ ) {
if (bRenderZ) {
// write depth as color
if( bRenderZToCol0 ) {
if( bOutputZ )
if (bRenderZToCol0) {
if (bOutputZ )
WRITE(p, "ocol0.xyz = frac(float3(256.0f*256.0f, 256.0f, 1.0f) * depth);\n");
else
WRITE(p, "ocol0.xyz = frac(float3(256.0f*256.0f, 256.0f, 1.0f) * uv%d.w);\n", ztexcoord);
WRITE(p, "ocol0.w = prev.w;\n");
}
else {
if( bOutputZ )
if (bOutputZ)
WRITE(p, "ocol1 = frac(float4(256.0f*256.0f, 256.0f, 1.0f, 0.0f) * depth);\n");
else
WRITE(p, "ocol1 = frac(float4(256.0f*256.0f, 256.0f, 1.0f, 0.0f) * uv%d.w);\n", ztexcoord);
}
}
WRITE(p,"}\n");
WRITE(p, "}\n");
return text;
}
@ -470,10 +470,10 @@ void WriteStage(char *&p, int n, u32 texture_mask)
bHasIndStage = true;
int texmap = bpmem.tevorders[n/2].getEnable(n&1) ? bpmem.tevorders[n/2].getTexMap(n&1) : bpmem.tevindref.getTexMap(bpmem.tevind[n].bt);
if( bpmem.tevind[n].bs != ITBA_OFF ) {
if (bpmem.tevind[n].bs != ITBA_OFF) {
// write the bump alpha
if( bpmem.tevind[n].fmt == ITF_8 ) {
if (bpmem.tevind[n].fmt == ITF_8) {
WRITE(p, "alphabump = indtex%d.%s %s;\n", bpmem.tevind[n].bt,
tevIndAlphaSel[bpmem.tevind[n].bs], tevIndAlphaScale[bpmem.tevind[n].fmt]);
}
@ -481,7 +481,7 @@ void WriteStage(char *&p, int n, u32 texture_mask)
// donkopunchstania: really bad way to do this
// cannot always use fract because fract(1.0) is 0.0 when it needs to be 1.0
// omitting fract seems to work as well
WRITE(p, "if( indtex%d.%s >= 1.0f )\n", bpmem.tevind[n].bt,
WRITE(p, "if (indtex%d.%s >= 1.0f )\n", bpmem.tevind[n].bt,
tevIndAlphaSel[bpmem.tevind[n].bs]);
WRITE(p, " alphabump = 1.0f;\n");
WRITE(p, "else\n");
@ -493,21 +493,21 @@ void WriteStage(char *&p, int n, u32 texture_mask)
// bias
WRITE(p, "float3 indtevcrd%d = indtex%d;\n", n, bpmem.tevind[n].bt);
WRITE(p, "indtevcrd%d.xy *= %s;\n", n, tevIndFmtScale[bpmem.tevind[n].fmt]);
if( bpmem.tevind[n].bias != ITB_NONE )
if (bpmem.tevind[n].bias != ITB_NONE )
WRITE(p, "indtevcrd%d.%s += %s;\n", n, tevIndBiasField[bpmem.tevind[n].bias], tevIndBiasAdd[bpmem.tevind[n].fmt]);
// multiply by offset matrix and scale
if( bpmem.tevind[n].mid != 0 ) {
if( bpmem.tevind[n].mid <= 3 ) {
if (bpmem.tevind[n].mid != 0) {
if (bpmem.tevind[n].mid <= 3) {
int mtxidx = 2*(bpmem.tevind[n].mid-1);
WRITE(p, "float2 indtevtrans%d = float2(dot("I_INDTEXMTX"[%d].xyz, indtevcrd%d), dot("I_INDTEXMTX"[%d].xyz, indtevcrd%d));\n",
n, mtxidx, n, mtxidx+1, n);
}
else if( bpmem.tevind[n].mid <= 5 ) { // s matrix
else if (bpmem.tevind[n].mid <= 5) { // s matrix
int mtxidx = 2*(bpmem.tevind[n].mid-5);
WRITE(p, "float2 indtevtrans%d = "I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.xx;\n", n, mtxidx, texcoord, n);
}
else if( bpmem.tevind[n].mid <= 9 ) { // t matrix
else if (bpmem.tevind[n].mid <= 9) { // t matrix
int mtxidx = 2*(bpmem.tevind[n].mid-9);
WRITE(p, "float2 indtevtrans%d = "I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.yy;\n", n, mtxidx, texcoord, n);
}
@ -522,12 +522,12 @@ void WriteStage(char *&p, int n, u32 texture_mask)
}
// wrapping
if( !bpmem.tevorders[n/2].getEnable(n&1) || (texture_mask & (1<<texmap)) ) {
if (!bpmem.tevorders[n/2].getEnable(n&1) || (texture_mask & (1<<texmap))) {
// non pow2
if( bpmem.tevind[n].sw != ITW_OFF || bpmem.tevind[n].tw != ITW_OFF ) {
if( bpmem.tevind[n].sw == ITW_0 ) {
if( bpmem.tevind[n].tw == ITW_0 ) {
if (bpmem.tevind[n].sw != ITW_OFF || bpmem.tevind[n].tw != ITW_OFF) {
if (bpmem.tevind[n].sw == ITW_0) {
if (bpmem.tevind[n].tw == ITW_0) {
// zero out completely
WRITE(p, "wrappedcoord = float2(0.0f,0.0f);\n");
}
@ -536,7 +536,7 @@ void WriteStage(char *&p, int n, u32 texture_mask)
"wrappedcoord.y = 0;\n", texcoord, tevIndWrapStart[bpmem.tevind[n].sw], texmap, texmap, tevIndWrapStart[bpmem.tevind[n].sw]);
}
}
else if( bpmem.tevind[n].tw == ITW_0 ) {
else if (bpmem.tevind[n].tw == ITW_0) {
WRITE(p, "wrappedcoord.y = fmod( (uv%d.y+%s)*"I_TEXDIMS"[%d].y*"I_TEXDIMS"[%d].w, %s);\n"
"wrappedcoord.x = 0;\n", texcoord, tevIndWrapStart[bpmem.tevind[n].tw], texmap, texmap, tevIndWrapStart[bpmem.tevind[n].tw]);
}
@ -555,9 +555,9 @@ void WriteStage(char *&p, int n, u32 texture_mask)
WRITE(p, "indtevtrans%d.xy *= "I_TEXDIMS"[%d].xy * "I_TEXDIMS"[%d].zw;\n", n, texmap, texmap);
// mult by bitdepth / tex dimensions
if( bpmem.tevind[n].sw != ITW_OFF || bpmem.tevind[n].tw != ITW_OFF ) {
if( bpmem.tevind[n].sw == ITW_0 ) {
if( bpmem.tevind[n].tw == ITW_0 ) {
if (bpmem.tevind[n].sw != ITW_OFF || bpmem.tevind[n].tw != ITW_OFF) {
if (bpmem.tevind[n].sw == ITW_0) {
if (bpmem.tevind[n].tw == ITW_0) {
// zero out completely
WRITE(p, "wrappedcoord = float2(0.0f,0.0f);\n");
}
@ -566,7 +566,7 @@ void WriteStage(char *&p, int n, u32 texture_mask)
"wrappedcoord.y = 0;\n", texmap, texcoord, tevIndWrapStart[bpmem.tevind[n].sw], texmap, tevIndWrapStart[bpmem.tevind[n].sw]);
}
}
else if( bpmem.tevind[n].tw == ITW_0 ) {
else if (bpmem.tevind[n].tw == ITW_0) {
WRITE(p, "wrappedcoord.y = "I_TEXDIMS"[%d].y * fmod( uv%d.y+%s, "I_TEXDIMS"[%d].w*%s);\n"
"wrappedcoord.x = 0;\n", texmap, texcoord, tevIndWrapStart[bpmem.tevind[n].tw], texmap, tevIndWrapStart[bpmem.tevind[n].tw]);
}
@ -582,25 +582,25 @@ void WriteStage(char *&p, int n, u32 texture_mask)
}
}
if( bpmem.tevind[n].fb_addprev ) {
if (bpmem.tevind[n].fb_addprev) {
// add previous tevcoord
if( texfun == XF_TEXPROJ_STQ ) {
WRITE(p,"tevcoord.xy += wrappedcoord/uv%d.z + indtevtrans%d;\n", texcoord, n);
//WRITE(p,"tevcoord.z += uv%d.z;\n", texcoord);
if (texfun == XF_TEXPROJ_STQ) {
WRITE(p, "tevcoord.xy += wrappedcoord/uv%d.z + indtevtrans%d;\n", texcoord, n);
//WRITE(p, "tevcoord.z += uv%d.z;\n", texcoord);
}
else {
WRITE(p,"tevcoord.xy += wrappedcoord + indtevtrans%d;\n", n);
WRITE(p, "tevcoord.xy += wrappedcoord + indtevtrans%d;\n", n);
}
}
else {
WRITE(p,"tevcoord.xy = wrappedcoord/uv%d.z + indtevtrans%d;\n", texcoord, n);
//if( texfun == XF_TEXPROJ_STQ )
// WRITE(p,"tevcoord.z = uv%d.z;\n", texcoord);
WRITE(p, "tevcoord.xy = wrappedcoord/uv%d.z + indtevtrans%d;\n", texcoord, n);
//if (texfun == XF_TEXPROJ_STQ )
// WRITE(p, "tevcoord.z = uv%d.z;\n", texcoord);
}
}
WRITE(p,"rastemp=%s.%s;\n",tevRasTable[bpmem.tevorders[n/2].getColorChan(n&1)],rasswap);
WRITE(p, "rastemp=%s.%s;\n",tevRasTable[bpmem.tevorders[n/2].getColorChan(n&1)],rasswap);
if (bpmem.tevorders[n/2].getEnable(n&1)) {
int texmap = bpmem.tevorders[n/2].getTexMap(n&1);
@ -612,41 +612,41 @@ void WriteStage(char *&p, int n, u32 texture_mask)
OurTexCoord = texcoord;
else
OurTexCoord = 0;
if( texture_mask & (1<<texmap) ) {
if (texture_mask & (1<<texmap)) {
// nonpow2
if( texfun == XF_TEXPROJ_STQ )
WRITE(p,"tevcoord.xy = uv%d.xy / uv%d.z;\n", texcoord, OurTexCoord);
if (texfun == XF_TEXPROJ_STQ )
WRITE(p, "tevcoord.xy = uv%d.xy / uv%d.z;\n", texcoord, OurTexCoord);
else
WRITE(p,"tevcoord.xy = uv%d.xy;\n", OurTexCoord);
WRITE(p, "tevcoord.xy = uv%d.xy;\n", OurTexCoord);
WrapNonPow2Tex(p, "tevcoord", texmap, texture_mask);
}
else {
if( texfun == XF_TEXPROJ_STQ )
WRITE(p,"tevcoord.xy = "I_TEXDIMS"[%d].xy * uv%d.xy / uv%d.z;\n", texmap, OurTexCoord , OurTexCoord );
if (texfun == XF_TEXPROJ_STQ )
WRITE(p, "tevcoord.xy = "I_TEXDIMS"[%d].xy * uv%d.xy / uv%d.z;\n", texmap, OurTexCoord , OurTexCoord );
else
WRITE(p,"tevcoord.xy = "I_TEXDIMS"[%d].xy * uv%d.xy;\n", texmap, OurTexCoord);
WRITE(p, "tevcoord.xy = "I_TEXDIMS"[%d].xy * uv%d.xy;\n", texmap, OurTexCoord);
}
}
else if( texture_mask & (1<<texmap) ) {
else if (texture_mask & (1<<texmap)) {
// if non pow 2, have to manually repeat
//WrapNonPow2Tex(p, "tevcoord", texmap);
bool bwraps = !!(texture_mask & (0x100<<texmap));
bool bwrapt = !!(texture_mask & (0x10000<<texmap));
if( bwraps || bwrapt ) {
if (bwraps || bwrapt) {
const char* field = bwraps ? (bwrapt ? "xy" : "x") : "y";
WRITE(p, "tevcoord.%s = fmod(tevcoord.%s+32*"I_TEXDIMS"[%d].%s,"I_TEXDIMS"[%d].%s);\n", field, field, texmap, field, texmap, field);
}
}
if( texture_mask & (1<<texmap) )
WRITE(p,"textemp=texRECT(samp%d,tevcoord.xy).%s;\n", texmap, texswap);
if (texture_mask & (1<<texmap) )
WRITE(p, "textemp=texRECT(samp%d,tevcoord.xy).%s;\n", texmap, texswap);
else
WRITE(p,"textemp=tex2D(samp%d,tevcoord.xy).%s;\n", texmap, texswap);
WRITE(p, "textemp=tex2D(samp%d,tevcoord.xy).%s;\n", texmap, texswap);
}
else
WRITE(p,"textemp=float4(1,1,1,1);\n");
WRITE(p, "textemp=float4(1,1,1,1);\n");
int kc = bpmem.tevksel[n/2].getKC(n&1);
int ka = bpmem.tevksel[n/2].getKA(n&1);
@ -656,59 +656,59 @@ void WriteStage(char *&p, int n, u32 texture_mask)
bool bCKonst = cc.a == TEVCOLORARG_KONST || cc.b == TEVCOLORARG_KONST || cc.c == TEVCOLORARG_KONST || cc.d == TEVCOLORARG_KONST;
bool bAKonst = ac.a == TEVALPHAARG_KONST || ac.b == TEVALPHAARG_KONST || ac.c == TEVALPHAARG_KONST || ac.d == TEVALPHAARG_KONST;
if( bCKonst || bAKonst )
WRITE(p,"konsttemp=float4(%s,%s);\n",tevKSelTableC[kc],tevKSelTableA[ka]);
if (bCKonst || bAKonst )
WRITE(p, "konsttemp=float4(%s,%s);\n",tevKSelTableC[kc],tevKSelTableA[ka]);
WRITE(p,"%s= ", tevCOutputTable[cc.dest]);
WRITE(p, "%s= ", tevCOutputTable[cc.dest]);
// combine the color channel
if (cc.bias != 3) { // if not compare
//normal color combiner goes here
WRITE(p," %s*(%s%s",tevScaleTable[cc.shift],tevCInputTable[cc.d],tevOpTable[cc.op]);
WRITE(p,"lerp(%s,%s,%s) %s);\n",
tevCInputTable[cc.a],tevCInputTable[cc.b],
tevCInputTable[cc.c],tevBiasTable[cc.bias]);
WRITE(p, " %s*(%s%s",tevScaleTable[cc.shift],tevCInputTable[cc.d],tevOpTable[cc.op]);
WRITE(p, "lerp(%s,%s,%s) %s);\n",
tevCInputTable[cc.a], tevCInputTable[cc.b],
tevCInputTable[cc.c], tevBiasTable[cc.bias]);
}
else {
int cmp = (cc.shift<<1)|cc.op|8; // comparemode stored here
switch(cmp) {
case TEVCMP_R8_GT:
case TEVCMP_RGB8_GT: // per component compares
WRITE(p," %s + ((%s.%s > %s.%s) ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d],tevCInputTable2[cc.a], cmp==TEVCMP_R8_GT?"r":"rgb", tevCInputTable2[cc.b], cmp==TEVCMP_R8_GT?"r":"rgb", tevCInputTable[cc.c]);
WRITE(p, " %s + ((%s.%s > %s.%s) ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d], tevCInputTable2[cc.a], cmp==TEVCMP_R8_GT?"r":"rgb", tevCInputTable2[cc.b], cmp==TEVCMP_R8_GT?"r":"rgb", tevCInputTable[cc.c]);
break;
case TEVCMP_R8_EQ:
case TEVCMP_RGB8_EQ:
WRITE(p," %s + (abs(%s.r - %s.r)<%f ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d],tevCInputTable2[cc.a], tevCInputTable2[cc.b],epsilon,tevCInputTable[cc.c]);
WRITE(p, " %s + (abs(%s.r - %s.r)<%f ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d], tevCInputTable2[cc.a], tevCInputTable2[cc.b], epsilon8bit, tevCInputTable[cc.c]);
break;
case TEVCMP_GR16_GT: // 16 bit compares: 255*g+r (probably used for ztextures, so make sure in ztextures, g is the most significant byte)
case TEVCMP_BGR24_GT: // 24 bit compares: 255*255*b+255*g+r
WRITE(p," %s + (( dot(%s.rgb-%s.rgb, comp%s) > 0) ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d],tevCInputTable2[cc.a], tevCInputTable2[cc.b], cmp==TEVCMP_GR16_GT?"16":"24", tevCInputTable[cc.c]);
WRITE(p, " %s + (( dot(%s.rgb-%s.rgb, comp%s) > 0) ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d], tevCInputTable2[cc.a], tevCInputTable2[cc.b], cmp==TEVCMP_GR16_GT?"16":"24", tevCInputTable[cc.c]);
break;
case TEVCMP_GR16_EQ:
case TEVCMP_BGR24_EQ:
WRITE(p," %s + (abs(dot(%s.rgb - %s.rgb, comp%s))<%f ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d],tevCInputTable2[cc.a], tevCInputTable2[cc.b],cmp==TEVCMP_GR16_GT?"16":"24",epsilon,tevCInputTable[cc.c]);
WRITE(p, " %s + (abs(dot(%s.rgb - %s.rgb, comp%s))<%f ? %s : float3(0.0f,0.0f,0.0f));\n",
tevCInputTable[cc.d], tevCInputTable2[cc.a], tevCInputTable2[cc.b], cmp==TEVCMP_GR16_GT?"16":"24", epsilon8bit, tevCInputTable[cc.c]);
break;
default:
WRITE(p,"float3(0.0f,0.0f,0.0f);\n");
WRITE(p, "float3(0.0f,0.0f,0.0f);\n");
break;
}
}
if( cc.clamp )
if (cc.clamp)
WRITE(p, "%s = clamp(%s,0.0f,1.0f);\n", tevCOutputTable[cc.dest],tevCOutputTable[cc.dest]);
// combine the alpha channel
WRITE(p,"%s= ", tevAOutputTable[ac.dest]);
WRITE(p, "%s= ", tevAOutputTable[ac.dest]);
if (ac.bias != 3) { // if not compare
//normal alpha combiner goes here
WRITE(p," %s*(%s%s",tevScaleTable[ac.shift],tevAInputTable[ac.d],tevOpTable[ac.op]);
WRITE(p,"lerp(%s,%s,%s) %s)\n",
WRITE(p, " %s*(%s%s",tevScaleTable[ac.shift],tevAInputTable[ac.d],tevOpTable[ac.op]);
WRITE(p, "lerp(%s,%s,%s) %s)\n",
tevAInputTable[ac.a],tevAInputTable[ac.b],
tevAInputTable[ac.c],tevBiasTable[ac.bias]);
}
@ -718,34 +718,34 @@ void WriteStage(char *&p, int n, u32 texture_mask)
switch(cmp) {
case TEVCMP_R8_GT:
case TEVCMP_A8_GT:
WRITE(p," %s + ((%s.%s > %s.%s) ? %s : 0)\n",
WRITE(p, " %s + ((%s.%s > %s.%s) ? %s : 0)\n",
tevAInputTable[ac.d],tevAInputTable2[ac.a], cmp==TEVCMP_R8_GT?"r":"a", tevAInputTable2[ac.b], cmp==TEVCMP_R8_GT?"r":"a", tevAInputTable[ac.c]);
break;
case TEVCMP_R8_EQ:
case TEVCMP_A8_EQ:
WRITE(p," %s + (abs(%s.r - %s.r)<%f ? %s : 0)\n",
tevAInputTable[ac.d],tevAInputTable2[ac.a], tevAInputTable2[ac.b],epsilon,tevAInputTable[ac.c]);
WRITE(p, " %s + (abs(%s.r - %s.r)<%f ? %s : 0)\n",
tevAInputTable[ac.d],tevAInputTable2[ac.a], tevAInputTable2[ac.b],epsilon8bit,tevAInputTable[ac.c]);
break;
case TEVCMP_GR16_GT: // 16 bit compares: 255*g+r (probably used for ztextures, so make sure in ztextures, g is the most significant byte)
case TEVCMP_BGR24_GT: // 24 bit compares: 255*255*b+255*g+r
WRITE(p," %s + (( dot(%s.rgb-%s.rgb, comp%s) > 0) ? %s : 0)\n",
WRITE(p, " %s + (( dot(%s.rgb-%s.rgb, comp%s) > 0) ? %s : 0)\n",
tevAInputTable[ac.d],tevAInputTable2[ac.a], tevAInputTable2[ac.b], cmp==TEVCMP_GR16_GT?"16":"24", tevAInputTable[ac.c]);
break;
case TEVCMP_GR16_EQ:
case TEVCMP_BGR24_EQ:
WRITE(p," %s + (abs(dot(%s.rgb - %s.rgb, comp%s))<%f ? %s : 0)\n",
tevAInputTable[ac.d],tevAInputTable2[ac.a], tevAInputTable2[ac.b],cmp==TEVCMP_GR16_GT?"16":"24",epsilon,tevAInputTable[ac.c]);
WRITE(p, " %s + (abs(dot(%s.rgb - %s.rgb, comp%s))<%f ? %s : 0)\n",
tevAInputTable[ac.d],tevAInputTable2[ac.a], tevAInputTable2[ac.b],cmp==TEVCMP_GR16_GT?"16":"24",epsilon8bit,tevAInputTable[ac.c]);
break;
default:
WRITE(p,"0)\n");
WRITE(p, "0)\n");
break;
}
}
WRITE(p,";\n");
WRITE(p, ";\n");
if( ac.clamp )
if (ac.clamp)
WRITE(p, "%s = clamp(%s,0.0f,1.0f);\n", tevAOutputTable[ac.dest],tevAOutputTable[ac.dest]);
WRITE(p, "\n");
}
@ -756,14 +756,14 @@ void WrapNonPow2Tex(char* &p, const char* var, int texmap, u32 texture_mask)
bool bwraps = !!(texture_mask & (0x100<<texmap));
bool bwrapt = !!(texture_mask & (0x10000<<texmap));
if( bwraps || bwrapt ) {
if (bwraps || bwrapt) {
const char* field = bwraps ? (bwrapt ? "xy" : "x") : "y";
const char* wrapfield = bwraps ? (bwrapt ? "zw" : "z") : "w";
WRITE(p, "%s.%s = "I_TEXDIMS"[%d].%s*frac(%s.%s*"I_TEXDIMS"[%d].%s+32);\n", var, field, texmap, field, var, field, texmap, wrapfield);
if( !bwraps )
if (!bwraps )
WRITE(p, "%s.x *= "I_TEXDIMS"[%d].x * "I_TEXDIMS"[%d].z;\n", var, texmap, texmap);
if( !bwrapt )
if (!bwrapt )
WRITE(p, "%s.y *= "I_TEXDIMS"[%d].y * "I_TEXDIMS"[%d].w;\n", var, texmap, texmap);
}
else {
@ -774,14 +774,14 @@ void WrapNonPow2Tex(char* &p, const char* var, int texmap, u32 texture_mask)
void WriteAlphaCompare(char *&p, int num, int comp)
{
switch(comp) {
case ALPHACMP_ALWAYS: WRITE(p,"(false)"); break;
case ALPHACMP_NEVER: WRITE(p,"(true)"); break;
case ALPHACMP_LEQUAL: WRITE(p,"(prev.a > %s)",alphaRef[num]); break;
case ALPHACMP_LESS: WRITE(p,"(prev.a >= %s - %f)",alphaRef[num],epsilon*0.5f);break;
case ALPHACMP_GEQUAL: WRITE(p,"(prev.a < %s)",alphaRef[num]); break;
case ALPHACMP_GREATER: WRITE(p,"(prev.a <= %s + %f)",alphaRef[num],epsilon*0.5f);break;
case ALPHACMP_EQUAL: WRITE(p,"(abs(prev.a-%s)>%f)",alphaRef[num],epsilon*2); break;
case ALPHACMP_NEQUAL: WRITE(p,"(abs(prev.a-%s)<%f)",alphaRef[num],epsilon*2); break;
case ALPHACMP_ALWAYS: WRITE(p, "(false)"); break;
case ALPHACMP_NEVER: WRITE(p, "(true)"); break;
case ALPHACMP_LEQUAL: WRITE(p, "(prev.a > %s)",alphaRef[num]); break;
case ALPHACMP_LESS: WRITE(p, "(prev.a >= %s - %f)",alphaRef[num],epsilon8bit*0.5f);break;
case ALPHACMP_GEQUAL: WRITE(p, "(prev.a < %s)",alphaRef[num]); break;
case ALPHACMP_GREATER: WRITE(p, "(prev.a <= %s + %f)",alphaRef[num],epsilon8bit*0.5f);break;
case ALPHACMP_EQUAL: WRITE(p, "(abs(prev.a-%s)>%f)",alphaRef[num],epsilon8bit*2); break;
case ALPHACMP_NEQUAL: WRITE(p, "(abs(prev.a-%s)<%f)",alphaRef[num],epsilon8bit*2); break;
}
}
@ -808,13 +808,13 @@ bool WriteAlphaTest(char *&p)
break;
case 2: // xor
if ( (comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_NEVER) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_ALWAYS) ) return true;
if ( (comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_ALWAYS) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_NEVER) ) {
if ( (comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_ALWAYS) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_NEVER)) {
WRITE(p, "discard;\n");
return false;
}
break;
case 3: // xnor
if ( (comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_NEVER) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_ALWAYS) ) {
if ( (comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_NEVER) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_ALWAYS)) {
WRITE(p, "discard;\n");
return false;
}

150
Source/Plugins/Plugin_VideoOGL/Src/PixelShaderManager.cpp
View File

@ -18,6 +18,9 @@
#include "Globals.h"
#include "Profiler.h"
#include <Cg/cg.h>
#include <Cg/cgGL.h>
#include <cmath>
#include "Common.h"
@ -65,7 +68,7 @@ void PixelShaderMngr::Init()
s_nIndTexMtxChanged = 15;
s_bAlphaChanged = s_bZBiasChanged = s_bIndTexScaleChanged = true;
GL_REPORT_ERRORD();
for(int i = 0; i < 8; ++i) maptocoord[i] = -1;
for (int i = 0; i < 8; ++i) maptocoord[i] = -1;
maptocoord_mask = 0;
memset(lastRGBAfull, 0, sizeof(lastRGBAfull));
@ -87,14 +90,13 @@ void PixelShaderMngr::Init()
"DP4 R1.y, R0, program.env[%d];\n"
"ADD result.color, R1, program.env[%d];\n"
"END\n", C_COLORMATRIX+3, C_COLORMATRIX+2, C_COLORMATRIX, C_COLORMATRIX+1, C_COLORMATRIX+4);
glGenProgramsARB( 1, &s_ColorMatrixProgram );
glBindProgramARB( GL_FRAGMENT_PROGRAM_ARB, s_ColorMatrixProgram );
glProgramStringARB( GL_FRAGMENT_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, (GLsizei)strlen(pmatrixprog), pmatrixprog);
glGenProgramsARB(1, &s_ColorMatrixProgram);
glBindProgramARB(GL_FRAGMENT_PROGRAM_ARB, s_ColorMatrixProgram);
glProgramStringARB(GL_FRAGMENT_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, (GLsizei)strlen(pmatrixprog), pmatrixprog);
GLenum err=GL_NO_ERROR;
GLenum err = GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
if (err != GL_NO_ERROR) {
ERROR_LOG("Failed to create color matrix fragment program\n");
glDeleteProgramsARB(1, &s_ColorMatrixProgram);
s_ColorMatrixProgram = 0;
@ -136,7 +138,7 @@ FRAGMENTSHADER* PixelShaderMngr::GetShader()
Renderer::GetRenderMode() != Renderer::RM_Normal);
#ifdef _DEBUG
if( g_Config.iLog & CONF_SAVESHADERS && code ) {
if (g_Config.iLog & CONF_SAVESHADERS && code) {
static int counter = 0;
char szTemp[MAX_PATH];
sprintf(szTemp, "%s/ps_%04i.txt", g_Config.texDumpPath, counter++);
@ -163,7 +165,7 @@ FRAGMENTSHADER* PixelShaderMngr::GetShader()
void PixelShaderMngr::Cleanup()
{
PSCache::iterator iter = pshaders.begin();
while(iter != pshaders.end()) {
while (iter != pshaders.end()) {
PSCacheEntry &entry = iter->second;
if (entry.frameCount < frameCount - 200) {
entry.Destroy();
@ -196,16 +198,16 @@ bool PixelShaderMngr::CompilePixelShader(FRAGMENTSHADER& ps, const char* pstrpro
char* pcompiledprog = (char*)cgGetProgramString(tempprog, CG_COMPILED_PROGRAM);
char* plocal = strstr(pcompiledprog, "program.local");
while( plocal != NULL ) {
while ( plocal != NULL) {
const char* penv = " program.env";
memcpy(plocal, penv, 13);
plocal = strstr(plocal+13, "program.local");
}
if( Renderer::IsUsingATIDrawBuffers() ) {
if (Renderer::IsUsingATIDrawBuffers()) {
// sometimes compilation can use ARB_draw_buffers, which would fail for ATI cards
char* poptions = strstr(pcompiledprog, "ARB_draw_buffers");
if( poptions != NULL ) {
if (poptions != NULL) {
poptions[0] = 'A';
poptions[1] = 'T';
poptions[2] = 'I';
@ -220,7 +222,7 @@ bool PixelShaderMngr::CompilePixelShader(FRAGMENTSHADER& ps, const char* pstrpro
GLenum err = GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
if (err != GL_NO_ERROR) {
ERROR_LOG(pstrprogram);
ERROR_LOG(pcompiledprog);
}
@ -236,12 +238,12 @@ bool PixelShaderMngr::CompilePixelShader(FRAGMENTSHADER& ps, const char* pstrpro
void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
{
for(int i = 0; i < 2; ++i) {
if( s_nColorsChanged[i] ) {
for (int i = 0; i < 2; ++i) {
if (s_nColorsChanged[i]) {
int baseind = i?C_KCOLORS:C_COLORS;
for(int j = 0; j < 4; ++j) {
if( s_nColorsChanged[i] & (1<<j) ) {
for (int j = 0; j < 4; ++j) {
if (s_nColorsChanged[i] & (1<<j)) {
SetPSConstant4fv(baseind+j, &lastRGBAfull[i][j][0]);
}
}
@ -250,8 +252,8 @@ void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
}
u32 newmask = 0;
for(u32 i = 0; i < (u32)bpmem.genMode.numtevstages+1; ++i) {
if( bpmem.tevorders[i/2].getEnable(i&1) ) {
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages+1; ++i) {
if (bpmem.tevorders[i/2].getEnable(i&1)) {
int texmap = bpmem.tevorders[i/2].getTexMap(i&1);
maptocoord[texmap] = bpmem.tevorders[i/2].getTexCoord(i&1);
newmask |= 1<<texmap;
@ -259,10 +261,10 @@ void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
}
}
if( maptocoord_mask != newmask ) {
if (maptocoord_mask != newmask) {
//u32 changes = maptocoord_mask ^ newmask;
for(int i = 0; i < 8; ++i) {
if( newmask&(1<<i) ) {
for (int i = 0; i < 8; ++i) {
if (newmask&(1<<i)) {
SetTexDimsChanged(i);
}
else {
@ -272,24 +274,24 @@ void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
maptocoord_mask = newmask;
}
if( s_nTexDimsChanged ) {
for(int i = 0; i < 8; ++i) {
if( s_nTexDimsChanged & (1<<i) ) {
if (s_nTexDimsChanged) {
for (int i = 0; i < 8; ++i) {
if (s_nTexDimsChanged & (1<<i)) {
SetPSTextureDims(i);
}
}
s_nTexDimsChanged = 0;
}
if( s_bAlphaChanged ) {
if (s_bAlphaChanged) {
SetPSConstant4f(C_ALPHA, (lastAlpha&0xff)/255.0f, ((lastAlpha>>8)&0xff)/255.0f, 0, ((lastAlpha>>16)&0xff)/255.0f);
}
if( s_bZBiasChanged ) {
if (s_bZBiasChanged) {
u32 bits;
float ffrac = 255.0f/256.0f;
float ftemp[4];
switch(bpmem.ztex2.type) {
switch (bpmem.ztex2.type) {
case 0:
bits = 8;
ftemp[0] = ffrac/(256.0f*256.0f); ftemp[1] = ffrac/256.0f; ftemp[2] = ffrac; ftemp[3] = 0;
@ -309,10 +311,10 @@ void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
}
// indirect incoming texture scales, update all!
if( s_bIndTexScaleChanged ) {
if (s_bIndTexScaleChanged) {
float f[8];
for(u32 i = 0; i < bpmem.genMode.numindstages; ++i) {
for (u32 i = 0; i < bpmem.genMode.numindstages; ++i) {
int srctexmap = bpmem.tevindref.getTexMap(i);
int texcoord = bpmem.tevindref.getTexCoord(i);
TCoordInfo& tc = bpmem.texcoords[texcoord];
@ -325,17 +327,19 @@ void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
SetPSConstant4fv(C_INDTEXSCALE, f);
if( bpmem.genMode.numindstages > 2 )
if (bpmem.genMode.numindstages > 2)
SetPSConstant4fv(C_INDTEXSCALE+1, &f[4]);
s_bIndTexScaleChanged = false;
}
if( s_nIndTexMtxChanged ) {
for(int i = 0; i < 3; ++i) {
if( s_nIndTexMtxChanged & (1<<i) ) {
int scale = ((u32)bpmem.indmtx[i].col0.s0<<0)|((u32)bpmem.indmtx[i].col1.s1<<2)|((u32)bpmem.indmtx[i].col2.s2<<4);
float fscale = powf(2.0f,(float)(scale-17)) / 1024.0f;
if (s_nIndTexMtxChanged) {
for (int i = 0; i < 3; ++i) {
if (s_nIndTexMtxChanged & (1 << i)) {
int scale = ((u32)bpmem.indmtx[i].col0.s0 << 0) |
((u32)bpmem.indmtx[i].col1.s1 << 2) |
((u32)bpmem.indmtx[i].col2.s2 << 4);
float fscale = powf(2.0f,(float)(scale - 17)) / 1024.0f;
// xyz - static matrix
//TODO w - dynamic matrix scale / 256...... somehow / 4 works better
@ -357,8 +361,8 @@ void PixelShaderMngr::SetPSTextureDims(int texid)
{
float fdims[4];
if( s_texturemask & (1<<texid) ) {
if( maptocoord[texid] >= 0 ) {
if (s_texturemask & (1<<texid)) {
if (maptocoord[texid] >= 0) {
TCoordInfo& tc = bpmem.texcoords[maptocoord[texid]];
fdims[0] = (float)(lastTexDims[texid]&0xffff);
fdims[1] = (float)((lastTexDims[texid]>>16)&0xfff);
@ -373,7 +377,7 @@ void PixelShaderMngr::SetPSTextureDims(int texid)
}
}
else {
if( maptocoord[texid] >= 0 ) {
if (maptocoord[texid] >= 0) {
TCoordInfo& tc = bpmem.texcoords[maptocoord[texid]];
fdims[0] = (float)(tc.s.scale_minus_1+1)/(float)(lastTexDims[texid]&0xffff);
fdims[1] = (float)(tc.t.scale_minus_1+1)/(float)((lastTexDims[texid]>>16)&0xfff);
@ -394,20 +398,22 @@ void PixelShaderMngr::SetPSTextureDims(int texid)
void PixelShaderMngr::SetColorChanged(int type, int num)
{
int r=bpmem.tevregs[num].low.a, a=bpmem.tevregs[num].low.b;
int b=bpmem.tevregs[num].high.a, g=bpmem.tevregs[num].high.b;
float* pf = &lastRGBAfull[type][num][0];
pf[0] = (float)r/255.0f;
pf[1] = (float)g/255.0f;
pf[2] = (float)b/255.0f;
pf[3] = (float)a/255.0f;
int r = bpmem.tevregs[num].low.a;
int a = bpmem.tevregs[num].low.b;
int b = bpmem.tevregs[num].high.a;
int g = bpmem.tevregs[num].high.b;
float *pf = &lastRGBAfull[type][num][0];
pf[0] = (float)r / 255.0f;
pf[1] = (float)g / 255.0f;
pf[2] = (float)b / 255.0f;
pf[3] = (float)a / 255.0f;
s_nColorsChanged[type] |= 1 << num;
PRIM_LOG("pixel %scolor%d: %f %f %f %f\n", type?"k":"", num, pf[0], pf[1], pf[2], pf[3]);
}
void PixelShaderMngr::SetAlpha(const AlphaFunc& alpha)
{
if( (alpha.hex&0xffff) != lastAlpha ) {
if ((alpha.hex&0xffff) != lastAlpha) {
lastAlpha = (lastAlpha&~0xffff)|(alpha.hex&0xffff);
s_bAlphaChanged = true;
}
@ -415,7 +421,7 @@ void PixelShaderMngr::SetAlpha(const AlphaFunc& alpha)
void PixelShaderMngr::SetDestAlpha(const ConstantAlpha& alpha)
{
if( alpha.alpha != (lastAlpha>>16) ) {
if (alpha.alpha != (lastAlpha>>16)) {
lastAlpha = (lastAlpha&~0xff0000)|((alpha.hex&0xff)<<16);
s_bAlphaChanged = true;
}
@ -423,8 +429,8 @@ void PixelShaderMngr::SetDestAlpha(const ConstantAlpha& alpha)
void PixelShaderMngr::SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt)
{
u32 wh = width|(height<<16)|(wraps<<28)|(wrapt<<30);
if( lastTexDims[texmapid] != wh ) {
u32 wh = width | (height<<16) | (wraps<<28) | (wrapt<<30);
if (lastTexDims[texmapid] != wh) {
lastTexDims[texmapid] = wh;
s_nTexDimsChanged |= 1<<texmapid;
}
@ -432,7 +438,7 @@ void PixelShaderMngr::SetTexDims(int texmapid, u32 width, u32 height, u32 wraps,
void PixelShaderMngr::SetZTetureBias(u32 bias)
{
if( lastZBias != bias ) {
if (lastZBias != bias) {
s_bZBiasChanged = true;
lastZBias = bias;
}
@ -475,11 +481,11 @@ void PixelShaderMngr::SetZTetureOpChanged()
void PixelShaderMngr::SetTexturesUsed(u32 nonpow2tex)
{
if( s_texturemask != nonpow2tex ) {
for(int i = 0; i < 8; ++i) {
if( nonpow2tex & (0x10101<<i) ) {
if (s_texturemask != nonpow2tex) {
for (int i = 0; i < 8; ++i) {
if (nonpow2tex & (0x10101 << i)) {
// this check was previously implicit, but should it be here?
if( s_nTexDimsChanged )
if (s_nTexDimsChanged )
s_nTexDimsChanged |= 1<<i;
}
}
@ -490,8 +496,8 @@ void PixelShaderMngr::SetTexturesUsed(u32 nonpow2tex)
void PixelShaderMngr::SetTexDimsChanged(int texmapid)
{
// this check was previously implicit, but should it be here?
if( s_nTexDimsChanged )
s_nTexDimsChanged |= 1<<texmapid;
if (s_nTexDimsChanged )
s_nTexDimsChanged |= 1 << texmapid;
SetIndTexScaleChanged();
}
@ -513,10 +519,10 @@ GLuint PixelShaderMngr::GetColorMatrixProgram()
void PixelShaderMngr::GetPixelShaderId(PixelShaderMngr::PIXELSHADERUID& uid)
{
u32 projtexcoords = 0;
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages+1; i++) {
if( bpmem.tevorders[i/2].getEnable(i&1) ) {
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages + 1; i++) {
if (bpmem.tevorders[i/2].getEnable(i&1)) {
int texcoord = bpmem.tevorders[i/2].getTexCoord(i&1);
if( xfregs.texcoords[texcoord].texmtxinfo.projection )
if (xfregs.texcoords[texcoord].texmtxinfo.projection )
projtexcoords |= 1<<texcoord;
}
}
@ -530,14 +536,14 @@ void PixelShaderMngr::GetPixelShaderId(PixelShaderMngr::PIXELSHADERUID& uid)
s_curuid.values[0] = (s_curuid.values[0]&~0x0ff00000)|(projtexcoords<<20);
// swap table
for(int i = 0; i < 8; i += 2)
for (int i = 0; i < 8; i += 2)
((u8*)&uid.values[1])[i/2] = (bpmem.tevksel[i].hex&0xf)|((bpmem.tevksel[i+1].hex&0xf)<<4);
uid.values[2] = s_texturemask;
int hdr = 3;
u32* pcurvalue = &uid.values[hdr];
for(u32 i = 0; i < (u32)bpmem.genMode.numtevstages+1; ++i) {
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages+1; ++i) {
TevStageCombiner::ColorCombiner &cc = bpmem.combiners[i].colorC;
TevStageCombiner::AlphaCombiner &ac = bpmem.combiners[i].alphaC;
@ -551,40 +557,40 @@ void PixelShaderMngr::GetPixelShaderId(PixelShaderMngr::PIXELSHADERUID& uid)
pcurvalue+=2;
}
for(u32 i = 0; i < ((u32)bpmem.genMode.numtevstages+1)/2; ++i) {
for (u32 i = 0; i < ((u32)bpmem.genMode.numtevstages+1)/2; ++i) {
u32 val0, val1;
if( bpmem.tevorders[i].hex&0x40 ) val0 = bpmem.tevorders[i].hex&0x3ff;
if (bpmem.tevorders[i].hex&0x40 ) val0 = bpmem.tevorders[i].hex&0x3ff;
else val0 = bpmem.tevorders[i].hex&0x380;
if( bpmem.tevorders[i].hex&0x40000 ) val1 = (bpmem.tevorders[i].hex&0x3ff000)>>12;
if (bpmem.tevorders[i].hex&0x40000 ) val1 = (bpmem.tevorders[i].hex&0x3ff000)>>12;
else val1 = (bpmem.tevorders[i].hex&0x380000)>>12;
switch(i % 3) {
switch (i % 3) {
case 0: pcurvalue[0] = val0|(val1<<10); break;
case 1: pcurvalue[0] |= val0<<20; pcurvalue[1] = val1; pcurvalue++; break;
case 2: pcurvalue[1] |= (val0<<10)|(val1<<20); pcurvalue++; break;
}
}
if( (bpmem.genMode.numtevstages+1)&1 ) { // odd
if ((bpmem.genMode.numtevstages+1)&1) { // odd
u32 val0;
if( bpmem.tevorders[bpmem.genMode.numtevstages/2].hex&0x40 ) val0 = bpmem.tevorders[bpmem.genMode.numtevstages/2].hex&0x3ff;
if (bpmem.tevorders[bpmem.genMode.numtevstages/2].hex&0x40 ) val0 = bpmem.tevorders[bpmem.genMode.numtevstages/2].hex&0x3ff;
else val0 = bpmem.tevorders[bpmem.genMode.numtevstages/2].hex&0x380;
switch(bpmem.genMode.numtevstages % 3) {
switch (bpmem.genMode.numtevstages % 3) {
case 0: pcurvalue[0] = val0; break;
case 1: pcurvalue[0] |= val0<<20; break;
case 2: pcurvalue[1] |= (val0<<10); pcurvalue++; break;
}
}
if( (bpmem.genMode.numtevstages % 3) != 2 )
if ((bpmem.genMode.numtevstages % 3) != 2 )
++pcurvalue;
uid.tevstages = (u32)(pcurvalue-&uid.values[0]-hdr);
for(u32 i = 0; i < bpmem.genMode.numindstages; ++i) {
for (u32 i = 0; i < bpmem.genMode.numindstages; ++i) {
u32 val = bpmem.tevind[i].hex&0x1fffff; // 21 bits
switch(i%3) {
switch (i%3) {
case 0: pcurvalue[0] = val; break;
case 1: pcurvalue[0] |= val<<21; pcurvalue[1] = val>>11; ++pcurvalue; break;
case 2: pcurvalue[0] |= val<<10; ++pcurvalue; break;

14
Source/Plugins/Plugin_VideoOGL/Src/Render.cpp
View File

@ -18,12 +18,16 @@
#include "Globals.h"
#include <list>
#include <Cg/cg.h>
#include <Cg/cgGL.h>
#ifdef _WIN32
#include <mmsystem.h>
#endif
#include "GLInit.h"
#include "Profiler.h"
#include "ImageWrite.h"
#include "Render.h"
#include "OpcodeDecoding.h"
#include "BPStructs.h"
@ -67,6 +71,8 @@ static Renderer::RenderMode s_RenderMode = Renderer::RM_Normal;
static int s_nCurTarget = 0;
bool g_bBlendLogicOp = false;
float MValueX, MValueY; // Since it can Stretch to fit Window, we need two different multiplication values
int frameCount;
void HandleCgError(CGcontext ctx, CGerror err, void* appdata);
@ -371,13 +377,12 @@ void Renderer::ProcessMessages()
{
GLboolean wasEnabled = glIsEnabled(GL_BLEND);
if(!wasEnabled) glEnable(GL_BLEND);
if (!wasEnabled) glEnable(GL_BLEND);
if (s_listMsgs.size() > 0) {
int left = 25, top = 15;
list<MESSAGE>::iterator it = s_listMsgs.begin();
while( it != s_listMsgs.end() )
while (it != s_listMsgs.end())
{
int time_left = (int)(it->dwTimeStamp - timeGetTime());
int alpha = 255;
@ -388,7 +393,7 @@ void Renderer::ProcessMessages()
if(time_left<0) alpha=0;
}
alpha<<=24;
alpha <<= 24;
RenderText(it->str, left+1, top+1, 0x000000|alpha);
RenderText(it->str, left, top, 0xffff30|alpha);
@ -761,7 +766,6 @@ void Renderer::SwapBuffers()
//p+=sprintf(p,"Num strip joins: %i\n",stats.numJoins);
p+=sprintf(p,"Num primitives: %i\n",stats.thisFrame.numPrims);
p+=sprintf(p,"Num primitives (DL): %i\n",stats.thisFrame.numDLPrims);
p+=sprintf(p,"Num bad commands: %i%s\n",stats.thisFrame.numBadCommands,stats.thisFrame.numBadCommands?"!!!":"");
p+=sprintf(p,"Num XF loads: %i\n",stats.thisFrame.numXFLoads);
p+=sprintf(p,"Num XF loads (DL): %i\n",stats.thisFrame.numXFLoadsInDL);
p+=sprintf(p,"Num CP loads: %i\n",stats.thisFrame.numCPLoads);

6
Source/Plugins/Plugin_VideoOGL/Src/Render.h
View File

@ -20,9 +20,15 @@
#include "TextureMngr.h"
#include <Cg/cg.h>
#include <Cg/cgGL.h>
extern CGcontext g_cgcontext;
extern CGprofile g_cgvProf, g_cgfProf;
extern float MValueX, MValueY;
extern int frameCount;
class Renderer
{
static void FlushZBufferAlphaToTarget();

2
Source/Plugins/Plugin_VideoOGL/Src/SConscript
View File

@ -18,6 +18,8 @@ files = [
'rasterfont.cpp',
'Render.cpp',
'TextureMngr.cpp',
'ImageWrite.cpp',
'VertexManager.cpp',
'VertexLoader.cpp',
'VertexLoader_Normal.cpp',
'VertexShader.cpp',

5
Source/Plugins/Plugin_VideoOGL/Src/TextureMngr.cpp
View File

@ -30,6 +30,7 @@
#endif
#include "Profiler.h"
#include "ImageWrite.h"
#include "Render.h"
@ -161,8 +162,8 @@ void TextureMngr::Cleanup()
}
std::map<u32, DEPTHTARGET>::iterator itdepth = mapDepthTargets.begin();
while(itdepth != mapDepthTargets.end()) {
if( frameCount > 20 + itdepth->second.framecount) {
while (itdepth != mapDepthTargets.end()) {
if (frameCount > 20 + itdepth->second.framecount) {
#ifdef _WIN32
itdepth = mapDepthTargets.erase(itdepth);
#else

411
Source/Plugins/Plugin_VideoOGL/Src/VertexLoader.cpp
View File

@ -16,16 +16,19 @@
// http://code.google.com/p/dolphin-emu/
#include "Globals.h"
#include <fstream>
#include <assert.h>
#include "Common.h"
#include "ImageWrite.h"
#include "x64Emitter.h"
#include "ABI.h"
#include "Profiler.h"
#include "StringUtil.h"
#include "Render.h"
#include "VertexManager.h"
#include "VertexLoader.h"
#include "BPStructs.h"
#include "DataReader.h"
@ -38,17 +41,7 @@
#include <fstream>
#define MAX_BUFFER_SIZE 0x4000
// internal state for loading vertices
static u32 s_prevvbstride, s_prevcomponents; // previous state set
static u8 *s_pBaseBufferPointer = NULL;
static GLuint s_vboBuffers[0x40] = {0};
static int s_nCurVBOIndex = 0; // current free buffer
static GLenum s_prevprimitive = 0; // current primitive type
static vector< pair<int, int> > s_vStoredPrimitives; // every element, mode and count to be passed to glDrawArrays
static void (*fnSetupVertexPointers)() = NULL;
extern void (*fnSetupVertexPointers)();
//these don't need to be saved
static float posScale;
@ -62,11 +55,18 @@ static int colIndex;
#define inline
#endif
TVtxDesc VertexManager::s_GlobalVtxDesc;
float VertexManager::shiftLookup[32];
// ==============================================================================
// Direct
// ==============================================================================
static u8 s_curposmtx, s_curtexmtx[8];
static int s_texmtxwrite = 0, s_texmtxread = 0;
static u8 s_curposmtx;
static u8 s_curtexmtx[8];
static int s_texmtxwrite = 0;
static int s_texmtxread = 0;
void LOADERDECL PosMtx_ReadDirect_UByte(void* _p)
{
@ -150,7 +150,7 @@ int VertexLoader::ComputeVertexSize()
m_VtxDesc.Hex = VertexManager::GetVtxDesc().Hex;
}
if( fnSetupVertexPointers != NULL && fnSetupVertexPointers == (void (*)())(void*)m_compiledCode )
if (fnSetupVertexPointers != NULL && fnSetupVertexPointers == (void (*)())(void*)m_compiledCode)
VertexManager::Flush();
m_AttrDirty = 1;
@ -258,7 +258,6 @@ int VertexLoader::ComputeVertexSize()
return m_VertexSize;
}
// Note the use of CallCdeclFunction3I etc.
// This is a horrible hack that is necessary because Opengl32.dll is based way, way above the 32-bit address space
// that is within reach of a CALL, and just doing &fn gives us these high uncallable addresses. So we want to grab
@ -519,8 +518,8 @@ void VertexLoader::ProcessFormat()
ABI_RestoreStack(1 * 4);
#endif
#endif
if( m_components&(VB_HAS_TEXMTXIDX0<<i) ) {
if( tc[i] != NOT_PRESENT ) {
if (m_components & (VB_HAS_TEXMTXIDX0 << i)) {
if (tc[i] != NOT_PRESENT) {
CallCdeclFunction4_I(glTexCoordPointer, 3, GL_FLOAT, m_VBVertexStride, offset);
offset += 12;
}
@ -668,32 +667,31 @@ void VertexLoader::WriteCall(void (LOADERDECL *func)(void *))
void VertexLoader::RunVertices(int primitive, int count)
{
ComputeVertexSize(); // HACK for underruns in Super Monkey Ball etc. !!!!
DVSTARTPROFILE();
if( count <= 0 )
ComputeVertexSize(); // HACK for underruns in Super Monkey Ball etc. !!!! dirty handling must be wrong.
if (count <= 0)
return;
if( fnSetupVertexPointers != NULL && fnSetupVertexPointers != (void (*)())(void*)m_compiledCode )
if (fnSetupVertexPointers != NULL && fnSetupVertexPointers != (void (*)())(void*)m_compiledCode)
VertexManager::Flush();
if( bpmem.genMode.cullmode == 3 && primitive < 5)
if (bpmem.genMode.cullmode == 3 && primitive < 5)
{
// if cull mode is none, ignore triangles and quads
DataSkip(count*m_VertexSize);
return;
}
DVSTARTPROFILE();
ProcessFormat();
fnSetupVertexPointers = (void (*)())(void*)m_compiledCode;
if (s_prevcomponents != m_components) {
// Move this code into VertexManager?
if (VertexManager::s_prevcomponents != m_components) {
VertexManager::Flush();
// matrices
if ((m_components & VB_HAS_POSMTXIDX) != (s_prevcomponents&VB_HAS_POSMTXIDX)) {
if ((m_components & VB_HAS_POSMTXIDX) != (VertexManager::s_prevcomponents & VB_HAS_POSMTXIDX)) {
if (m_components & VB_HAS_POSMTXIDX)
glEnableVertexAttribArray(SHADER_POSMTX_ATTRIB);
else
@ -701,13 +699,13 @@ void VertexLoader::RunVertices(int primitive, int count)
}
// normals
if ((m_components & VB_HAS_NRM0) != (s_prevcomponents&VB_HAS_NRM0)) {
if ((m_components & VB_HAS_NRM0) != (VertexManager::s_prevcomponents & VB_HAS_NRM0)) {
if (m_components & VB_HAS_NRM0)
glEnableClientState(GL_NORMAL_ARRAY);
else
glDisableClientState(GL_NORMAL_ARRAY);
}
if ((m_components & VB_HAS_NRM1) != (s_prevcomponents&VB_HAS_NRM1)) {
if ((m_components & VB_HAS_NRM1) != (VertexManager::s_prevcomponents & VB_HAS_NRM1)) {
if (m_components & VB_HAS_NRM1) {
glEnableVertexAttribArray(SHADER_NORM1_ATTRIB);
glEnableVertexAttribArray(SHADER_NORM2_ATTRIB);
@ -719,29 +717,28 @@ void VertexLoader::RunVertices(int primitive, int count)
}
// color
for(int i = 0; i < 2; ++i) {
if ( (m_components & (VB_HAS_COL0 << i)) != (s_prevcomponents & (VB_HAS_COL0 << i)) ) {
for (int i = 0; i < 2; ++i) {
if ((m_components & (VB_HAS_COL0 << i)) != (VertexManager::s_prevcomponents & (VB_HAS_COL0 << i))) {
if (m_components & (VB_HAS_COL0 << 0))
glEnableClientState(i?GL_SECONDARY_COLOR_ARRAY:GL_COLOR_ARRAY);
glEnableClientState(i ? GL_SECONDARY_COLOR_ARRAY : GL_COLOR_ARRAY);
else
glDisableClientState(i?GL_SECONDARY_COLOR_ARRAY:GL_COLOR_ARRAY);
glDisableClientState(i ? GL_SECONDARY_COLOR_ARRAY : GL_COLOR_ARRAY);
}
}
// tex
for (int i = 0; i < 8; ++i) {
if ((m_components&(VB_HAS_UV0<<i)) != (s_prevcomponents&(VB_HAS_UV0<<i))) {
glClientActiveTexture(GL_TEXTURE0+i);
if (m_components&(VB_HAS_UV0<<i))
if ((m_components & (VB_HAS_UV0 << i)) != (VertexManager::s_prevcomponents & (VB_HAS_UV0 << i))) {
glClientActiveTexture(GL_TEXTURE0 + i);
if (m_components & (VB_HAS_UV0 << i))
glEnableClientState(GL_TEXTURE_COORD_ARRAY);
else
glDisableClientState(GL_TEXTURE_COORD_ARRAY);
}
}
s_prevcomponents = m_components;
s_prevvbstride = m_VBVertexStride;
VertexManager::s_prevcomponents = m_components;
VertexManager::s_prevvbstride = m_VBVertexStride;
}
PrepareRun();
@ -771,20 +768,20 @@ void VertexLoader::RunVertices(int primitive, int count)
}
int startv = 0, extraverts = 0;
for (int v = 0; v < count; v++) {
if( (v % granularity) == 0 ) {
for (int v = 0; v < count; v++)
{
if ((v % granularity) == 0)
{
if (VertexManager::GetRemainingSize() < granularity*m_VBVertexStride) {
u8* plastptr = VertexManager::s_pCurBufferPointer;
if( v-startv > 0 )
if (v-startv > 0)
VertexManager::AddVertices(primitive, v-startv+extraverts);
VertexManager::Flush();
switch( primitive ) {
// Why does this need to be so complicated?
switch (primitive) {
case 3: // triangle strip, copy last two vertices
// a little trick since we have to keep track of signs
if( v & 1 ) {
if (v & 1) {
memcpy_gc(VertexManager::s_pCurBufferPointer, plastptr-2*m_VBVertexStride, m_VBVertexStride);
memcpy_gc(VertexManager::s_pCurBufferPointer+m_VBVertexStride, plastptr-m_VBVertexStride*2, 2*m_VBVertexStride);
VertexManager::s_pCurBufferPointer += m_VBVertexStride*3;
@ -812,339 +809,19 @@ void VertexLoader::RunVertices(int primitive, int count)
extraverts = 0;
break;
}
startv = v;
}
}
tcIndex = 0;
colIndex = 0;
s_texmtxwrite = s_texmtxread = 0;
// int pred_size = m_VertexSize;
//int start = GetBufferPosition();
//if (!m_numPipelineStates)
// PanicAlert("trying to draw with no pipeline");
for (int i = 0; i < m_numPipelineStates; i++)
m_PipelineStates[i](&m_VtxAttr);
//int end = GetBufferPosition();
//if (end - start != pred_size) {
// std::string vtx_summary;
// vtx_summary += StringFromFormat("Nrm d:%i f:%i e:%i 3:%i", m_VtxDesc.Normal, m_VtxAttr.NormalFormat, m_VtxAttr.NormalElements, m_VtxAttr.NormalIndex3);
// PanicAlert((vtx_summary + "\nWTF %i %i").c_str(), end - start, pred_size);
//}
VertexManager::s_pCurBufferPointer += m_VBStridePad;
PRIM_LOG("\n");
}
if( startv < count )
if (startv < count)
VertexManager::AddVertices(primitive, count-startv+extraverts);
}
///////////////////
// VertexManager //
///////////////////
TVtxDesc VertexManager::s_GlobalVtxDesc;
u8* VertexManager::s_pCurBufferPointer=NULL;
float VertexManager::shiftLookup[32];
const GLenum c_primitiveType[8] =
{
GL_QUADS,
0, //nothing
GL_TRIANGLES,
GL_TRIANGLE_STRIP,
GL_TRIANGLE_FAN,
GL_LINES,
GL_LINE_STRIP,
GL_POINTS
};
bool VertexManager::Init()
{
Destroy();
s_GlobalVtxDesc.Hex = 0;
s_prevcomponents = 0;
s_prevvbstride = 12; // just pos
s_prevprimitive = 0;
s_pBaseBufferPointer = (u8*)AllocateMemoryPages(MAX_BUFFER_SIZE);
s_pCurBufferPointer = s_pBaseBufferPointer;
for (u32 i = 0; i < ARRAYSIZE(shiftLookup); i++)
shiftLookup[i] = 1.0f / float(1 << i);
s_nCurVBOIndex = 0;
glGenBuffers(ARRAYSIZE(s_vboBuffers), s_vboBuffers);
for (u32 i = 0; i < ARRAYSIZE(s_vboBuffers); ++i) {
glBindBuffer(GL_ARRAY_BUFFER, s_vboBuffers[i]);
glBufferData(GL_ARRAY_BUFFER, MAX_BUFFER_SIZE, NULL, GL_STREAM_DRAW);
}
glEnableClientState(GL_VERTEX_ARRAY);
fnSetupVertexPointers = NULL;
GL_REPORT_ERRORD();
return true;
}
void VertexManager::Destroy()
{
FreeMemoryPages(s_pBaseBufferPointer, MAX_BUFFER_SIZE); s_pBaseBufferPointer = s_pCurBufferPointer = NULL;
glDeleteBuffers(ARRAYSIZE(s_vboBuffers), s_vboBuffers);
memset(s_vboBuffers, 0, sizeof(s_vboBuffers));
s_vStoredPrimitives.resize(0);
s_nCurVBOIndex = 0;
ResetBuffer();
}
void VertexManager::ResetBuffer()
{
s_nCurVBOIndex = (s_nCurVBOIndex+1)%ARRAYSIZE(s_vboBuffers);
s_pCurBufferPointer = s_pBaseBufferPointer;
s_vStoredPrimitives.resize(0);
}
void VertexManager::ResetComponents()
{
s_prevcomponents = 0;
s_prevvbstride = 12; // just pos
s_prevprimitive = 0;
glDisableVertexAttribArray(SHADER_POSMTX_ATTRIB);
glDisableClientState(GL_NORMAL_ARRAY);
glDisableVertexAttribArray(SHADER_NORM1_ATTRIB);
glDisableVertexAttribArray(SHADER_NORM2_ATTRIB);
glDisableClientState(GL_COLOR_ARRAY);
glDisableClientState(GL_SECONDARY_COLOR_ARRAY);
for (int i = 0; i < 8; ++i) glDisableClientState(GL_TEXTURE_COORD_ARRAY);
}
int VertexManager::GetRemainingSize()
{
return MAX_BUFFER_SIZE - (int)(s_pCurBufferPointer-s_pBaseBufferPointer);
}
void VertexManager::AddVertices(int primitive, int numvertices)
{
_assert_( numvertices > 0 );
ADDSTAT(stats.thisFrame.numPrims, numvertices);
s_vStoredPrimitives.push_back(pair<int, int>(c_primitiveType[primitive], numvertices));
#ifdef _DEBUG
static const char *sprims[8] = {"quads", "nothing", "tris", "tstrip", "tfan", "lines", "lstrip", "points"};
PRIM_LOG("prim: %s, c=%d\n", sprims[primitive], numvertices);
#endif
}
void VertexManager::Flush()
{
if (s_vStoredPrimitives.size() == 0)
return;
_assert_( fnSetupVertexPointers != NULL );
_assert_( s_pCurBufferPointer != s_pBaseBufferPointer );
#ifdef _DEBUG
PRIM_LOG("frame%d:\ncomps=0x%x, texgen=%d, numchan=%d, dualtex=%d, ztex=%d, proj=%d, cole=%d, alpe=%d, ze=%d\n", g_Config.iSaveTargetId, s_prevcomponents, xfregs.numTexGens,
xfregs.nNumChans, (int)xfregs.bEnableDualTexTransform, bpmem.ztex2.op, VertexShaderMngr::rawProjection[6]==0,
bpmem.blendmode.colorupdate, bpmem.blendmode.alphaupdate, bpmem.zmode.updateenable);
for(int i = 0; i < xfregs.nNumChans; ++i) {
LitChannel* ch = &xfregs.colChans[i].color;
PRIM_LOG("colchan%d: matsrc=%d, light=0x%x, ambsrc=%d, diffunc=%d, attfunc=%d\n", i, ch->matsource, ch->GetFullLightMask(), ch->ambsource, ch->diffusefunc, ch->attnfunc);
ch = &xfregs.colChans[i].alpha;
PRIM_LOG("alpchan%d: matsrc=%d, light=0x%x, ambsrc=%d, diffunc=%d, attfunc=%d\n", i, ch->matsource, ch->GetFullLightMask(), ch->ambsource, ch->diffusefunc, ch->attnfunc);
}
for(int i = 0; i < xfregs.numTexGens; ++i) {
TexMtxInfo tinfo = xfregs.texcoords[i].texmtxinfo;
if( tinfo.texgentype != XF_TEXGEN_EMBOSS_MAP ) tinfo.hex &= 0x7ff;
if( tinfo.texgentype != XF_TEXGEN_REGULAR ) tinfo.projection = 0;
PRIM_LOG("txgen%d: proj=%d, input=%d, gentype=%d, srcrow=%d, embsrc=%d, emblght=%d, postmtx=%d, postnorm=%d\n",
i, tinfo.projection, tinfo.inputform, tinfo.texgentype, tinfo.sourcerow, tinfo.embosssourceshift, tinfo.embosslightshift,
xfregs.texcoords[i].postmtxinfo.index, xfregs.texcoords[i].postmtxinfo.normalize);
}
PRIM_LOG("pixel: tev=%d, ind=%d, texgen=%d, dstalpha=%d, alphafunc=0x%x\n", bpmem.genMode.numtevstages+1, bpmem.genMode.numindstages,
bpmem.genMode.numtexgens, (u32)bpmem.dstalpha.enable, (bpmem.alphaFunc.hex>>16)&0xff);
#endif
DVSTARTPROFILE();
GL_REPORT_ERRORD();
glBindBuffer(GL_ARRAY_BUFFER, s_vboBuffers[s_nCurVBOIndex]);
glBufferData(GL_ARRAY_BUFFER, s_pCurBufferPointer-s_pBaseBufferPointer, s_pBaseBufferPointer, GL_STREAM_DRAW);
GL_REPORT_ERRORD();
// setup the pointers
fnSetupVertexPointers();
GL_REPORT_ERRORD();
// set the textures
{
DVProfileFunc _pf("VertexManager::Flush:textures");
u32 usedtextures = 0;
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages+1; ++i) {
if( bpmem.tevorders[i/2].getEnable(i&1) )
usedtextures |= 1<<bpmem.tevorders[i/2].getTexMap(i&1);
}
if( bpmem.genMode.numindstages > 0 ) {
for(u32 i = 0; i < (u32)bpmem.genMode.numtevstages+1; ++i) {
if( bpmem.tevind[i].IsActive() && bpmem.tevind[i].bt < bpmem.genMode.numindstages ) {
usedtextures |= 1<<bpmem.tevindref.getTexMap(bpmem.tevind[i].bt);
}
}
}
u32 nonpow2tex = 0;
for (int i = 0; i < 8; i++) {
if (usedtextures&(1<<i)) {
glActiveTexture(GL_TEXTURE0+i);
FourTexUnits &tex = bpmem.tex[i>>2];
TextureMngr::TCacheEntry* tentry = TextureMngr::Load(i, (tex.texImage3[i&3].image_base/* & 0x1FFFFF*/) << 5,
tex.texImage0[i&3].width+1, tex.texImage0[i&3].height+1,
tex.texImage0[i&3].format, tex.texTlut[i&3].tmem_offset<<9, tex.texTlut[i&3].tlut_format);
if( tentry != NULL ) {
// texture loaded fine, set dims for pixel shader
if( tentry->isNonPow2 ) {
PixelShaderMngr::SetTexDims(i, tentry->w, tentry->h, tentry->mode.wrap_s, tentry->mode.wrap_t);
nonpow2tex |= 1<<i;
if( tentry->mode.wrap_s > 0 ) nonpow2tex |= 1<<(8+i);
if( tentry->mode.wrap_t > 0 ) nonpow2tex |= 1<<(16+i);
TextureMngr::EnableTexRECT(i);
}
// if texture is power of two, set to ones (since don't need scaling)
else
{
PixelShaderMngr::SetTexDims(i, tentry->w, tentry->h, 0, 0);
TextureMngr::EnableTex2D(i);
}
if( g_Config.iLog & CONF_PRIMLOG ) {
// save the textures
char strfile[255];
sprintf(strfile, "frames/tex%.3d_%d.tga", g_Config.iSaveTargetId, i);
SaveTexture(strfile, tentry->isNonPow2?GL_TEXTURE_RECTANGLE_ARB:GL_TEXTURE_2D, tentry->texture, tentry->w, tentry->h);
}
}
else {
ERROR_LOG("error loading tex\n");
TextureMngr::DisableStage(i); // disable since won't be used
}
}
else {
TextureMngr::DisableStage(i); // disable since won't be used
}
}
PixelShaderMngr::SetTexturesUsed(nonpow2tex);
}
FRAGMENTSHADER* ps = PixelShaderMngr::GetShader();
VERTEXSHADER* vs = VertexShaderMngr::GetShader(s_prevcomponents);
_assert_( ps != NULL && vs != NULL );
bool bRestoreBuffers = false;
if( Renderer::GetZBufferTarget() ) {
if( bpmem.zmode.updateenable ) {
if( !bpmem.blendmode.colorupdate ) {
Renderer::SetRenderMode(bpmem.blendmode.alphaupdate?Renderer::RM_ZBufferAlpha:Renderer::RM_ZBufferOnly);
}
}
else {
Renderer::SetRenderMode(Renderer::RM_Normal);
// remove temporarily
glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT);
bRestoreBuffers = true;
}
}
else
Renderer::SetRenderMode(Renderer::RM_Normal);
// set global constants
VertexShaderMngr::SetConstants(*vs);
PixelShaderMngr::SetConstants(*ps);
// finally bind
glBindProgramARB(GL_VERTEX_PROGRAM_ARB, vs->glprogid);
glBindProgramARB(GL_FRAGMENT_PROGRAM_ARB, ps->glprogid);
PRIM_LOG("\n");
int offset = 0;
vector< pair<int, int> >::iterator it;
for (it = s_vStoredPrimitives.begin(); it != s_vStoredPrimitives.end(); ++it) {
glDrawArrays(it->first, offset, it->second);
offset += it->second;
}
#ifdef _DEBUG
if( g_Config.iLog & CONF_PRIMLOG ) {
// save the shaders
char strfile[255];
sprintf(strfile, "frames/ps%.3d.txt", g_Config.iSaveTargetId);
std::ofstream fps(strfile);
fps << ps->strprog.c_str();
sprintf(strfile, "frames/vs%.3d.txt", g_Config.iSaveTargetId);
ofstream fvs(strfile);
fvs << vs->strprog.c_str();
}
if( g_Config.iLog & CONF_SAVETARGETS ) {
char str[128];
sprintf(str, "frames/targ%.3d.tga", g_Config.iSaveTargetId);
Renderer::SaveRenderTarget(str, 0);
}
#endif
g_Config.iSaveTargetId++;
GL_REPORT_ERRORD();
if( bRestoreBuffers ) {
GLenum s_drawbuffers[2] = {GL_COLOR_ATTACHMENT0_EXT, GL_COLOR_ATTACHMENT1_EXT};
glDrawBuffers(2, s_drawbuffers);
SetColorMask();
}
ResetBuffer();
}
void VertexManager::LoadCPReg(u32 SubCmd, u32 Value)
{
switch (SubCmd & 0xF0)
{
case 0x30:
VertexShaderMngr::SetTexMatrixChangedA(Value);
break;
case 0x40:
VertexShaderMngr::SetTexMatrixChangedB(Value);
break;
case 0x50:
s_GlobalVtxDesc.Hex &= ~0x1FFFF; // keep the Upper bits
s_GlobalVtxDesc.Hex |= Value;
break;
case 0x60:
s_GlobalVtxDesc.Hex &= 0x1FFFF; // keep the lower 17Bits
s_GlobalVtxDesc.Hex |= (u64)Value << 17;
break;
case 0x70: g_VertexLoaders[SubCmd & 7].SetVAT_group0(Value); _assert_((SubCmd & 0x0F) < 8); break;
case 0x80: g_VertexLoaders[SubCmd & 7].SetVAT_group1(Value); _assert_((SubCmd & 0x0F) < 8); break;
case 0x90: g_VertexLoaders[SubCmd & 7].SetVAT_group2(Value); _assert_((SubCmd & 0x0F) < 8); break;
case 0xA0: arraybases[SubCmd & 0xF] = Value & 0xFFFFFFFF; break;
case 0xB0: arraystrides[SubCmd & 0xF] = Value & 0xFF; break;
}
}

45
Source/Plugins/Plugin_VideoOGL/Src/VertexLoader.h
View File

@ -32,7 +32,8 @@ using namespace std;
#define LOADERDECL __cdecl
typedef void (LOADERDECL *TPipelineFunction)(void*);
/// Use to manage loading and setting vertex buffer data for OpenGL
// There are 8 of these. Most games only use the first, and just reconfigure it all the time
// as needed, unfortunately.
class VertexLoader
{
public:
@ -97,10 +98,10 @@ private:
//common for all loaders
TVtxDesc m_VtxDesc;
// seup the pipeline with this vertex fmt
void SetupColor(int num, int _iMode, int _iFormat, int _iElements);
void SetupTexCoord(int num, int _iMode, int _iFormat, int _iElements, int _iFrac);
// The 3 possible values (0, 1, 2) should be documented here.
int m_AttrDirty;
public:
@ -124,7 +125,7 @@ public:
void SetVAT_group0(u32 _group0)
{
if ((m_group0.Hex&~0x3e0001f0) != (_group0&~0x3e0001f0)) {
if ((m_group0.Hex & ~0x3e0001f0) != (_group0 & ~0x3e0001f0)) {
m_AttrDirty = 2;
}
m_group0.Hex = _group0;
@ -147,7 +148,7 @@ public:
void SetVAT_group1(u32 _group1)
{
if ((m_group1.Hex&~0x7c3e1f0) != (_group1&~0x7c3e1f0)) {
if ((m_group1.Hex & ~0x7c3e1f0) != (_group1 & ~0x7c3e1f0)) {
m_AttrDirty = 2;
}
m_group1.Hex = _group1;
@ -168,7 +169,7 @@ public:
void SetVAT_group2(u32 _group2)
{
if ((m_group2.Hex&~0xf87c3e1f) != (_group2&~0xf87c3e1f)) {
if ((m_group2.Hex & ~0xf87c3e1f) != (_group2 & ~0xf87c3e1f)) {
m_AttrDirty = 2;
}
m_group2.Hex = _group2;
@ -186,38 +187,6 @@ public:
};
};
/// Methods to manage and cache the global state of vertex streams and flushing streams
/// Also handles processing the CP registers
class VertexManager
{
static TVtxDesc s_GlobalVtxDesc;
public:
enum Collection
{
C_NOTHING=0,
C_TRIANGLES=1,
C_LINES=2,
C_POINTS=3
};
static bool Init();
static void Destroy();
static void ResetBuffer();
static void ResetComponents();
static void AddVertices(int primitive, int numvertices);
static void Flush(); // flushes the current buffer
static int GetRemainingSize();
static TVtxDesc &GetVtxDesc() {return s_GlobalVtxDesc; }
static void LoadCPReg(u32 SubCmd, u32 Value);
static u8* s_pCurBufferPointer;
static float shiftLookup[32];
};
extern VertexLoader g_VertexLoaders[8];
#endif

8
Source/Plugins/Plugin_VideoOGL/Src/VertexLoader_Normal.cpp
View File

@ -15,11 +15,9 @@
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
//__________________________________________________________________________________________________
// F|RES 2003-2005
//
#include "Globals.h"
#include "VertexLoader.h"
#include "VertexManager.h"
#include "VertexLoader_Normal.h"
#define LOG_NORM8() PRIM_LOG("norm: %f %f %f, ", ((s8*)VertexManager::s_pCurBufferPointer)[-3]/127.0f, ((s8*)VertexManager::s_pCurBufferPointer)[-2]/127.0f, ((s8*)VertexManager::s_pCurBufferPointer)[-1]/127.0f);
@ -30,9 +28,7 @@ u8 VertexLoader_Normal::m_sizeTable[NUM_NRM_TYPE][NUM_NRM_FORMAT][NUM_NRM_EL
TPipelineFunction VertexLoader_Normal::m_funcTable[NUM_NRM_TYPE][NUM_NRM_FORMAT][NUM_NRM_ELEMENTS];
bool VertexLoader_Normal::index3;
// __________________________________________________________________________________________________
// Init
//
void VertexLoader_Normal::Init(void)
{
// size table

325
Source/Plugins/Plugin_VideoOGL/Src/VertexManager.cpp Normal file
View File

@ -0,0 +1,325 @@
#include "Globals.h"
#include "MemoryUtil.h"
#include "Profiler.h"
#include "Render.h"
#include "ImageWrite.h"
#include "BPMemory.h"
#include "TextureMngr.h"
#include "PixelShaderManager.h"
#include "VertexShaderManager.h"
#include "VertexLoader.h"
#include "VertexManager.h"
#define MAX_BUFFER_SIZE 0x4000
static GLuint s_vboBuffers[0x40] = {0};
static int s_nCurVBOIndex = 0; // current free buffer
static GLenum s_prevprimitive = 0; // current primitive type
static u8 *s_pBaseBufferPointer = NULL;
static vector< pair<int, int> > s_vStoredPrimitives; // every element, mode and count to be passed to glDrawArrays
u8* VertexManager::s_pCurBufferPointer = NULL;
u32 VertexManager::s_prevvbstride;
u32 VertexManager::s_prevcomponents; // previous state set
const GLenum c_primitiveType[8] =
{
GL_QUADS,
0, //nothing
GL_TRIANGLES,
GL_TRIANGLE_STRIP,
GL_TRIANGLE_FAN,
GL_LINES,
GL_LINE_STRIP,
GL_POINTS
};
// internal state for loading vertices
void (*fnSetupVertexPointers)() = NULL;
bool VertexManager::Init()
{
Destroy();
s_GlobalVtxDesc.Hex = 0;
s_prevcomponents = 0;
s_prevvbstride = 12; // just pos
s_prevprimitive = 0;
s_pBaseBufferPointer = (u8*)AllocateMemoryPages(MAX_BUFFER_SIZE);
s_pCurBufferPointer = s_pBaseBufferPointer;
for (u32 i = 0; i < ARRAYSIZE(shiftLookup); i++)
shiftLookup[i] = 1.0f / float(1 << i);
s_nCurVBOIndex = 0;
glGenBuffers(ARRAYSIZE(s_vboBuffers), s_vboBuffers);
for (u32 i = 0; i < ARRAYSIZE(s_vboBuffers); ++i) {
glBindBuffer(GL_ARRAY_BUFFER, s_vboBuffers[i]);
glBufferData(GL_ARRAY_BUFFER, MAX_BUFFER_SIZE, NULL, GL_STREAM_DRAW);
}
glEnableClientState(GL_VERTEX_ARRAY);
fnSetupVertexPointers = NULL;
GL_REPORT_ERRORD();
return true;
}
void VertexManager::Destroy()
{
FreeMemoryPages(s_pBaseBufferPointer, MAX_BUFFER_SIZE); s_pBaseBufferPointer = s_pCurBufferPointer = NULL;
glDeleteBuffers(ARRAYSIZE(s_vboBuffers), s_vboBuffers);
memset(s_vboBuffers, 0, sizeof(s_vboBuffers));
s_vStoredPrimitives.resize(0);
s_nCurVBOIndex = 0;
ResetBuffer();
}
void VertexManager::ResetBuffer()
{
s_nCurVBOIndex = (s_nCurVBOIndex+1)%ARRAYSIZE(s_vboBuffers);
s_pCurBufferPointer = s_pBaseBufferPointer;
s_vStoredPrimitives.resize(0);
}
void VertexManager::ResetComponents()
{
s_prevcomponents = 0;
s_prevvbstride = 12; // just pos
s_prevprimitive = 0;
glDisableVertexAttribArray(SHADER_POSMTX_ATTRIB);
glDisableClientState(GL_NORMAL_ARRAY);
glDisableVertexAttribArray(SHADER_NORM1_ATTRIB);
glDisableVertexAttribArray(SHADER_NORM2_ATTRIB);
glDisableClientState(GL_COLOR_ARRAY);
glDisableClientState(GL_SECONDARY_COLOR_ARRAY);
for (int i = 0; i < 8; ++i) glDisableClientState(GL_TEXTURE_COORD_ARRAY);
}
int VertexManager::GetRemainingSize()
{
return MAX_BUFFER_SIZE - (int)(s_pCurBufferPointer-s_pBaseBufferPointer);
}
void VertexManager::AddVertices(int primitive, int numvertices)
{
_assert_( numvertices > 0 );
ADDSTAT(stats.thisFrame.numPrims, numvertices);
s_vStoredPrimitives.push_back(pair<int, int>(c_primitiveType[primitive], numvertices));
#ifdef _DEBUG
static const char *sprims[8] = {"quads", "nothing", "tris", "tstrip", "tfan", "lines", "lstrip", "points"};
PRIM_LOG("prim: %s, c=%d\n", sprims[primitive], numvertices);
#endif
}
void VertexManager::Flush()
{
if (s_vStoredPrimitives.size() == 0)
return;
_assert_( fnSetupVertexPointers != NULL );
_assert_( s_pCurBufferPointer != s_pBaseBufferPointer );
#ifdef _DEBUG
PRIM_LOG("frame%d:\ncomps=0x%x, texgen=%d, numchan=%d, dualtex=%d, ztex=%d, proj=%d, cole=%d, alpe=%d, ze=%d\n", g_Config.iSaveTargetId, s_prevcomponents, xfregs.numTexGens,
xfregs.nNumChans, (int)xfregs.bEnableDualTexTransform, bpmem.ztex2.op, VertexShaderMngr::rawProjection[6]==0,
bpmem.blendmode.colorupdate, bpmem.blendmode.alphaupdate, bpmem.zmode.updateenable);
for (int i = 0; i < xfregs.nNumChans; ++i) {
LitChannel* ch = &xfregs.colChans[i].color;
PRIM_LOG("colchan%d: matsrc=%d, light=0x%x, ambsrc=%d, diffunc=%d, attfunc=%d\n", i, ch->matsource, ch->GetFullLightMask(), ch->ambsource, ch->diffusefunc, ch->attnfunc);
ch = &xfregs.colChans[i].alpha;
PRIM_LOG("alpchan%d: matsrc=%d, light=0x%x, ambsrc=%d, diffunc=%d, attfunc=%d\n", i, ch->matsource, ch->GetFullLightMask(), ch->ambsource, ch->diffusefunc, ch->attnfunc);
}
for (int i = 0; i < xfregs.numTexGens; ++i) {
TexMtxInfo tinfo = xfregs.texcoords[i].texmtxinfo;
if (tinfo.texgentype != XF_TEXGEN_EMBOSS_MAP ) tinfo.hex &= 0x7ff;
if (tinfo.texgentype != XF_TEXGEN_REGULAR ) tinfo.projection = 0;
PRIM_LOG("txgen%d: proj=%d, input=%d, gentype=%d, srcrow=%d, embsrc=%d, emblght=%d, postmtx=%d, postnorm=%d\n",
i, tinfo.projection, tinfo.inputform, tinfo.texgentype, tinfo.sourcerow, tinfo.embosssourceshift, tinfo.embosslightshift,
xfregs.texcoords[i].postmtxinfo.index, xfregs.texcoords[i].postmtxinfo.normalize);
}
PRIM_LOG("pixel: tev=%d, ind=%d, texgen=%d, dstalpha=%d, alphafunc=0x%x\n", bpmem.genMode.numtevstages+1, bpmem.genMode.numindstages,
bpmem.genMode.numtexgens, (u32)bpmem.dstalpha.enable, (bpmem.alphaFunc.hex>>16)&0xff);
#endif
DVSTARTPROFILE();
GL_REPORT_ERRORD();
glBindBuffer(GL_ARRAY_BUFFER, s_vboBuffers[s_nCurVBOIndex]);
glBufferData(GL_ARRAY_BUFFER, s_pCurBufferPointer-s_pBaseBufferPointer, s_pBaseBufferPointer, GL_STREAM_DRAW);
GL_REPORT_ERRORD();
// setup the pointers
fnSetupVertexPointers();
GL_REPORT_ERRORD();
// set the textures
{
DVProfileFunc _pf("VertexManager::Flush:textures");
u32 usedtextures = 0;
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages + 1; ++i) {
if (bpmem.tevorders[i/2].getEnable(i & 1))
usedtextures |= 1<<bpmem.tevorders[i/2].getTexMap(i & 1);
}
if (bpmem.genMode.numindstages > 0) {
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages + 1; ++i) {
if (bpmem.tevind[i].IsActive() && bpmem.tevind[i].bt < bpmem.genMode.numindstages) {
usedtextures |= 1<<bpmem.tevindref.getTexMap(bpmem.tevind[i].bt);
}
}
}
u32 nonpow2tex = 0;
for (int i = 0; i < 8; i++) {
if (usedtextures & (1 << i)) {
glActiveTexture(GL_TEXTURE0+i);
FourTexUnits &tex = bpmem.tex[i>>2];
TextureMngr::TCacheEntry* tentry = TextureMngr::Load(i, (tex.texImage3[i&3].image_base/* & 0x1FFFFF*/) << 5,
tex.texImage0[i&3].width+1, tex.texImage0[i&3].height+1,
tex.texImage0[i&3].format, tex.texTlut[i&3].tmem_offset<<9, tex.texTlut[i&3].tlut_format);
if (tentry != NULL) {
// texture loaded fine, set dims for pixel shader
if (tentry->isNonPow2) {
PixelShaderMngr::SetTexDims(i, tentry->w, tentry->h, tentry->mode.wrap_s, tentry->mode.wrap_t);
nonpow2tex |= 1<<i;
if (tentry->mode.wrap_s > 0 ) nonpow2tex |= 1<<(8+i);
if (tentry->mode.wrap_t > 0 ) nonpow2tex |= 1<<(16+i);
TextureMngr::EnableTexRECT(i);
}
// if texture is power of two, set to ones (since don't need scaling)
else
{
PixelShaderMngr::SetTexDims(i, tentry->w, tentry->h, 0, 0);
TextureMngr::EnableTex2D(i);
}
if (g_Config.iLog & CONF_PRIMLOG) {
// save the textures
char strfile[255];
sprintf(strfile, "frames/tex%.3d_%d.tga", g_Config.iSaveTargetId, i);
SaveTexture(strfile, tentry->isNonPow2?GL_TEXTURE_RECTANGLE_ARB:GL_TEXTURE_2D, tentry->texture, tentry->w, tentry->h);
}
}
else {
ERROR_LOG("error loading tex\n");
TextureMngr::DisableStage(i); // disable since won't be used
}
}
else {
TextureMngr::DisableStage(i); // disable since won't be used
}
}
PixelShaderMngr::SetTexturesUsed(nonpow2tex);
}
FRAGMENTSHADER* ps = PixelShaderMngr::GetShader();
VERTEXSHADER* vs = VertexShaderMngr::GetShader(s_prevcomponents);
_assert_( ps != NULL && vs != NULL );
bool bRestoreBuffers = false;
if (Renderer::GetZBufferTarget()) {
if (bpmem.zmode.updateenable) {
if (!bpmem.blendmode.colorupdate) {
Renderer::SetRenderMode(bpmem.blendmode.alphaupdate?Renderer::RM_ZBufferAlpha:Renderer::RM_ZBufferOnly);
}
}
else {
Renderer::SetRenderMode(Renderer::RM_Normal);
// remove temporarily
glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT);
bRestoreBuffers = true;
}
}
else
Renderer::SetRenderMode(Renderer::RM_Normal);
// set global constants
VertexShaderMngr::SetConstants(*vs);
PixelShaderMngr::SetConstants(*ps);
// finally bind
glBindProgramARB(GL_VERTEX_PROGRAM_ARB, vs->glprogid);
glBindProgramARB(GL_FRAGMENT_PROGRAM_ARB, ps->glprogid);
#ifdef _DEBUG
PRIM_LOG("\n");
#endif
int offset = 0;
vector< pair<int, int> >::iterator it;
for (it = s_vStoredPrimitives.begin(); it != s_vStoredPrimitives.end(); ++it) {
glDrawArrays(it->first, offset, it->second);
offset += it->second;
}
#ifdef _DEBUG
if (g_Config.iLog & CONF_PRIMLOG) {
// save the shaders
char strfile[255];
sprintf(strfile, "frames/ps%.3d.txt", g_Config.iSaveTargetId);
std::ofstream fps(strfile);
fps << ps->strprog.c_str();
sprintf(strfile, "frames/vs%.3d.txt", g_Config.iSaveTargetId);
ofstream fvs(strfile);
fvs << vs->strprog.c_str();
}
if (g_Config.iLog & CONF_SAVETARGETS) {
char str[128];
sprintf(str, "frames/targ%.3d.tga", g_Config.iSaveTargetId);
Renderer::SaveRenderTarget(str, 0);
}
#endif
g_Config.iSaveTargetId++;
GL_REPORT_ERRORD();
if (bRestoreBuffers) {
GLenum s_drawbuffers[2] = {GL_COLOR_ATTACHMENT0_EXT, GL_COLOR_ATTACHMENT1_EXT};
glDrawBuffers(2, s_drawbuffers);
SetColorMask();
}
ResetBuffer();
}
void VertexManager::LoadCPReg(u32 SubCmd, u32 Value)
{
switch (SubCmd & 0xF0)
{
case 0x30:
VertexShaderMngr::SetTexMatrixChangedA(Value);
break;
case 0x40:
VertexShaderMngr::SetTexMatrixChangedB(Value);
break;
case 0x50:
s_GlobalVtxDesc.Hex &= ~0x1FFFF; // keep the Upper bits
s_GlobalVtxDesc.Hex |= Value;
break;
case 0x60:
s_GlobalVtxDesc.Hex &= 0x1FFFF; // keep the lower 17Bits
s_GlobalVtxDesc.Hex |= (u64)Value << 17;
break;
case 0x70: g_VertexLoaders[SubCmd & 7].SetVAT_group0(Value); _assert_((SubCmd & 0x0F) < 8); break;
case 0x80: g_VertexLoaders[SubCmd & 7].SetVAT_group1(Value); _assert_((SubCmd & 0x0F) < 8); break;
case 0x90: g_VertexLoaders[SubCmd & 7].SetVAT_group2(Value); _assert_((SubCmd & 0x0F) < 8); break;
case 0xA0: arraybases[SubCmd & 0xF] = Value & 0xFFFFFFFF; break;
case 0xB0: arraystrides[SubCmd & 0xF] = Value & 0xFF; break;
}
}

60
Source/Plugins/Plugin_VideoOGL/Src/VertexManager.h Normal file
View File

@ -0,0 +1,60 @@
// Copyright (C) 2003-2008 Dolphin Project.
// 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, version 2.0.
// 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 2.0 for more details.
// A copy of the GPL 2.0 should have been included with the program.
// If not, see http://www.gnu.org/licenses/
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
#ifndef _VERTEXMANAGER_H
#define _VERTEXMANAGER_H
#include "CPMemory.h"
// Methods to manage and cache the global state of vertex streams and flushing streams
// Also handles processing the CP registers
class VertexManager
{
static TVtxDesc s_GlobalVtxDesc;
public:
enum Collection
{
C_NOTHING=0,
C_TRIANGLES=1,
C_LINES=2,
C_POINTS=3
};
static bool Init();
static void Destroy();
static void ResetBuffer();
static void ResetComponents();
static void AddVertices(int primitive, int numvertices);
static void Flush(); // flushes the current buffer
static int GetRemainingSize();
static TVtxDesc &GetVtxDesc() {return s_GlobalVtxDesc; }
static void LoadCPReg(u32 SubCmd, u32 Value);
// TODO - don't expose these like this.
static u32 s_prevvbstride;
static u32 s_prevcomponents; // previous state set
static u8* s_pCurBufferPointer;
static float shiftLookup[32];
};
#endif // _VERTEXMANAGER_H

162
Source/Plugins/Plugin_VideoOGL/Src/VertexShader.cpp
View File

@ -40,17 +40,17 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
_assert_( bpmem.genMode.numcolchans == xfregs.nNumChans);
u32 lightMask = 0;
if( xfregs.nNumChans > 0 )
lightMask |= xfregs.colChans[0].color.GetFullLightMask() | xfregs.colChans[0].alpha.GetFullLightMask();
if( xfregs.nNumChans > 1 )
if (xfregs.nNumChans > 0)
lightMask |= xfregs.colChans[0].color.GetFullLightMask() | xfregs.colChans[0].alpha.GetFullLightMask();
if (xfregs.nNumChans > 1)
lightMask |= xfregs.colChans[1].color.GetFullLightMask() | xfregs.colChans[1].alpha.GetFullLightMask();
bool bOutputZ = bpmem.ztex2.op==ZTEXTURE_ADD || has_zbuffer_target;
int ztexcoord = -1;
char *p = text;
WRITE(p,"//Vertex Shader: comp:%x, \n", components);
WRITE(p,"typedef struct {\n"
WRITE(p, "//Vertex Shader: comp:%x, \n", components);
WRITE(p, "typedef struct {\n"
" float4 T0, T1, T2;\n"
" float4 N0, N1, N2;\n"
"} s_"I_POSNORMALMATRIX";\n\n"
@ -89,25 +89,25 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
" float4 params;\n" // a, b, c, b_shift
"} s_"I_FOGPARAMS";\n\n");
WRITE(p,"struct VS_OUTPUT {\n");
WRITE(p," float4 pos : POSITION;\n");
WRITE(p," float4 colors[2] : COLOR0;\n");
WRITE(p, "struct VS_OUTPUT {\n");
WRITE(p, " float4 pos : POSITION;\n");
WRITE(p, " float4 colors[2] : COLOR0;\n");
// if outputting Z, embed the Z coordinate in the w component of a texture coordinate
// if number of tex gens occupies all the texture coordinates, use the last tex coord
// otherwise use the next available tex coord
for (int i = 0; i < xfregs.numTexGens; ++i) {
WRITE(p," float%d tex%d : TEXCOORD%d;\n", (i==(xfregs.numTexGens-1)&&bOutputZ)?4:3, i, i);
WRITE(p, " float%d tex%d : TEXCOORD%d;\n", (i==(xfregs.numTexGens-1)&&bOutputZ)?4:3, i, i);
}
if (bOutputZ && xfregs.numTexGens == 0) {
ztexcoord = 0;
WRITE(p," float4 tex%d : TEXCOORD%d;\n", ztexcoord, ztexcoord);
WRITE(p, " float4 tex%d : TEXCOORD%d;\n", ztexcoord, ztexcoord);
}
else if (bOutputZ)
ztexcoord = xfregs.numTexGens - 1;
WRITE(p,"};\n");
WRITE(p,"\n");
WRITE(p, "};\n");
WRITE(p, "\n");
// uniforms
@ -123,152 +123,152 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
WRITE(p, "uniform s_"I_PROJECTION" "I_PROJECTION" : register(c%d);\n", C_PROJECTION);
WRITE(p, "uniform s_"I_FOGPARAMS" "I_FOGPARAMS" : register(c%d);\n", C_FOGPARAMS);
WRITE(p,"VS_OUTPUT main(\n");
WRITE(p, "VS_OUTPUT main(\n");
// inputs
if (components & VertexLoader::VB_HAS_NRM0)
WRITE(p," float3 rawnorm0 : NORMAL,\n");
WRITE(p, " float3 rawnorm0 : NORMAL,\n");
if (components & VertexLoader::VB_HAS_NRM1)
WRITE(p," float3 rawnorm1 : ATTR%d,\n", SHADER_NORM1_ATTRIB);
WRITE(p, " float3 rawnorm1 : ATTR%d,\n", SHADER_NORM1_ATTRIB);
if (components & VertexLoader::VB_HAS_NRM2)
WRITE(p," float3 rawnorm2 : ATTR%d,\n", SHADER_NORM2_ATTRIB);
WRITE(p, " float3 rawnorm2 : ATTR%d,\n", SHADER_NORM2_ATTRIB);
if (components & VertexLoader::VB_HAS_COL0)
WRITE(p," float4 color0 : COLOR0,\n");
WRITE(p, " float4 color0 : COLOR0,\n");
if (components & VertexLoader::VB_HAS_COL1)
WRITE(p," float4 color1 : COLOR1,\n");
WRITE(p, " float4 color1 : COLOR1,\n");
for (int i = 0; i < 8; ++i) {
u32 hastexmtx = (components & (VertexLoader::VB_HAS_TEXMTXIDX0<<i));
if ( (components & (VertexLoader::VB_HAS_UV0<<i)) || hastexmtx )
WRITE(p," float%d tex%d : TEXCOORD%d,\n", hastexmtx ? 3 : 2, i,i);
if ((components & (VertexLoader::VB_HAS_UV0<<i)) || hastexmtx )
WRITE(p, " float%d tex%d : TEXCOORD%d,\n", hastexmtx ? 3 : 2, i,i);
}
if (components & VertexLoader::VB_HAS_POSMTXIDX)
WRITE(p, " half posmtx : ATTR%d,\n", SHADER_POSMTX_ATTRIB);
WRITE(p," float4 rawpos : POSITION) {\n");
WRITE(p, " float4 rawpos : POSITION) {\n");
WRITE(p, "VS_OUTPUT o;\n");
// transforms
if ( components & VertexLoader::VB_HAS_POSMTXIDX) {
if (components & VertexLoader::VB_HAS_POSMTXIDX) {
WRITE(p, "float4 pos = float4(dot("I_TRANSFORMMATRICES".T[posmtx].t, rawpos), dot("I_TRANSFORMMATRICES".T[posmtx+1].t, rawpos), dot("I_TRANSFORMMATRICES".T[posmtx+2].t, rawpos),1);\n");
if (components & VertexLoader::VB_HAS_NRMALL) {
WRITE(p, "int normidx = posmtx >= 32 ? (posmtx-32) : posmtx;\n");
WRITE(p,"float3 N0 = "I_NORMALMATRICES".T[normidx].t.xyz, N1 = "I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = "I_NORMALMATRICES".T[normidx+2].t.xyz;\n");
WRITE(p, "float3 N0 = "I_NORMALMATRICES".T[normidx].t.xyz, N1 = "I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = "I_NORMALMATRICES".T[normidx+2].t.xyz;\n");
}
if (components & VertexLoader::VB_HAS_NRM0)
WRITE(p,"half3 _norm0 = half3(dot(N0, rawnorm0), dot(N1, rawnorm0), dot(N2, rawnorm0));\n"
"half3 norm0 = normalize(_norm0);\n");
WRITE(p, "half3 _norm0 = half3(dot(N0, rawnorm0), dot(N1, rawnorm0), dot(N2, rawnorm0));\n"
"half3 norm0 = normalize(_norm0);\n");
if (components & VertexLoader::VB_HAS_NRM1)
WRITE(p,"half3 _norm1 = half3(dot(N0, rawnorm1), dot(N1, rawnorm1), dot(N2, rawnorm1));\n");
WRITE(p, "half3 _norm1 = half3(dot(N0, rawnorm1), dot(N1, rawnorm1), dot(N2, rawnorm1));\n");
//"half3 norm1 = normalize(_norm1);\n");
if (components & VertexLoader::VB_HAS_NRM2)
WRITE(p,"half3 _norm2 = half3(dot(N0, rawnorm2), dot(N1, rawnorm2), dot(N2, rawnorm2));\n");
WRITE(p, "half3 _norm2 = half3(dot(N0, rawnorm2), dot(N1, rawnorm2), dot(N2, rawnorm2));\n");
//"half3 norm2 = normalize(_norm2);\n");
}
else {
WRITE(p, "float4 pos = float4(dot("I_POSNORMALMATRIX".T0, rawpos), dot("I_POSNORMALMATRIX".T1, rawpos), dot("I_POSNORMALMATRIX".T2, rawpos), 1);\n");
if (components & VertexLoader::VB_HAS_NRM0)
WRITE(p,"half3 _norm0 = half3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm0));\n"
"half3 norm0 = normalize(_norm0);\n");
WRITE(p, "half3 _norm0 = half3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm0));\n"
"half3 norm0 = normalize(_norm0);\n");
if (components & VertexLoader::VB_HAS_NRM1)
WRITE(p,"half3 _norm1 = half3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n");
WRITE(p, "half3 _norm1 = half3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n");
//"half3 norm1 = normalize(_norm1);\n");
if (components & VertexLoader::VB_HAS_NRM2)
WRITE(p,"half3 _norm2 = half3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n");
WRITE(p, "half3 _norm2 = half3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n");
//"half3 norm2 = normalize(_norm2);\n");
}
if (!(components & VertexLoader::VB_HAS_NRM0))
WRITE(p,"half3 _norm0 = half3(0,0,0), norm0= half3(0,0,0);\n");
WRITE(p, "half3 _norm0 = half3(0,0,0), norm0= half3(0,0,0);\n");
WRITE(p,"o.pos = float4(dot("I_PROJECTION".T0, pos), dot("I_PROJECTION".T1, pos), dot("I_PROJECTION".T2, pos), dot("I_PROJECTION".T3, pos));\n");
WRITE(p, "o.pos = float4(dot("I_PROJECTION".T0, pos), dot("I_PROJECTION".T1, pos), dot("I_PROJECTION".T2, pos), dot("I_PROJECTION".T3, pos));\n");
WRITE(p, "half4 mat, lacc;\n"
"half3 ldir, h;\n"
"half dist, dist2, attn;\n");
// lights/colors
for (int j=0; j<xfregs.nNumChans; j++) {
for (int j = 0; j < xfregs.nNumChans; j++) {
// bool bColorAlphaSame = xfregs.colChans[j].color.hex == xfregs.colChans[j].alpha.hex; unused
const LitChannel& color = xfregs.colChans[j].color;
const LitChannel& alpha = xfregs.colChans[j].alpha;
WRITE(p,"{\n");
WRITE(p, "{\n");
if (color.matsource) {// from vertex
if (components & (VertexLoader::VB_HAS_COL0<<j) )
WRITE(p,"mat = color%d;\n", j);
else WRITE(p,"mat = half4(1,1,1,1);\n");
WRITE(p, "mat = color%d;\n", j);
else WRITE(p, "mat = half4(1,1,1,1);\n");
}
else // from color
WRITE(p,"mat = "I_MATERIALS".C%d;\n", j+2);
WRITE(p, "mat = "I_MATERIALS".C%d;\n", j+2);
if( color.enablelighting ) {
if (color.enablelighting) {
if (color.ambsource) {// from vertex
if (components & (VertexLoader::VB_HAS_COL0<<j) )
WRITE(p,"lacc = color%d;\n", j);
else WRITE(p,"lacc = half4(0.0f,0.0f,0.0f,0.0f);\n");
WRITE(p, "lacc = color%d;\n", j);
else WRITE(p, "lacc = half4(0.0f,0.0f,0.0f,0.0f);\n");
}
else // from color
WRITE(p,"lacc = "I_MATERIALS".C%d;\n", j);
WRITE(p, "lacc = "I_MATERIALS".C%d;\n", j);
}
// check if alpha is different
if (alpha.matsource != color.matsource) {
if (alpha.matsource ) {// from vertex
if (alpha.matsource) {// from vertex
if (components & (VertexLoader::VB_HAS_COL0<<j) )
WRITE(p,"mat.w = color%d.w;\n", j);
else WRITE(p,"mat.w = 1;\n");
WRITE(p, "mat.w = color%d.w;\n", j);
else WRITE(p, "mat.w = 1;\n");
}
else // from color
WRITE(p,"mat.w = "I_MATERIALS".C%d.w;\n", j+2);
WRITE(p, "mat.w = "I_MATERIALS".C%d.w;\n", j+2);
}
if (alpha.enablelighting && alpha.ambsource != color.ambsource) {
if (alpha.ambsource) {// from vertex
if (components & (VertexLoader::VB_HAS_COL0<<j) )
WRITE(p,"lacc.w = color%d.w;\n", j);
else WRITE(p,"lacc.w = 0;\n");
WRITE(p, "lacc.w = color%d.w;\n", j);
else WRITE(p, "lacc.w = 0;\n");
}
else // from color
WRITE(p,"lacc.w = "I_MATERIALS".C%d.w;\n", j);
WRITE(p, "lacc.w = "I_MATERIALS".C%d.w;\n", j);
}
if( color.enablelighting && alpha.enablelighting && (color.GetFullLightMask() != alpha.GetFullLightMask() || color.lightparams != alpha.lightparams) ) {
if (color.enablelighting && alpha.enablelighting && (color.GetFullLightMask() != alpha.GetFullLightMask() || color.lightparams != alpha.lightparams)) {
// both have lighting, except not using the same lights
int mask = 0; // holds already computed lights
if( color.lightparams == alpha.lightparams && (color.GetFullLightMask() & alpha.GetFullLightMask()) ) {
if (color.lightparams == alpha.lightparams && (color.GetFullLightMask() & alpha.GetFullLightMask())) {
// if lights are shared, compute those first
mask = color.GetFullLightMask() & alpha.GetFullLightMask();
for(int i = 0; i < 8; ++i) {
if( mask&(1<<i))
for (int i = 0; i < 8; ++i) {
if (mask&(1<<i))
p = GenerateLightShader(p, i, color, "lacc", 3);
}
}
// no shared lights
for(int i = 0; i < 8; ++i) {
if( !(mask&(1<<i)) && (color.GetFullLightMask() & (1<<i)) )
for (int i = 0; i < 8; ++i) {
if (!(mask&(1<<i)) && (color.GetFullLightMask() & (1<<i)) )
p = GenerateLightShader(p, i, color, "lacc", 1);
if( !(mask&(1<<i)) && (alpha.GetFullLightMask() & (1<<i)) )
if (!(mask&(1<<i)) && (alpha.GetFullLightMask() & (1<<i)) )
p = GenerateLightShader(p, i, alpha, "lacc", 2);
}
}
else if( color.enablelighting || alpha.enablelighting) {
else if (color.enablelighting || alpha.enablelighting) {
// either one is enabled
int coloralpha = (int)color.enablelighting|((int)alpha.enablelighting<<1);
for(int i = 0; i < 8; ++i) {
if( color.GetFullLightMask() & (1<<i) )
for (int i = 0; i < 8; ++i) {
if (color.GetFullLightMask() & (1<<i) )
p = GenerateLightShader(p, i, color.enablelighting?color:alpha, "lacc", coloralpha);
}
}
if (color.enablelighting != alpha.enablelighting) {
if( color.enablelighting )
if (color.enablelighting )
WRITE(p, "o.colors[%d].xyz = mat.xyz * clamp(lacc.xyz,float3(0.0f,0.0f,0.0f),float3(1.0f,1.0f,1.0f));\n"
"o.colors[%d].w = mat.w;\n", j, j);
else
@ -276,7 +276,7 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
"o.colors[%d].w = mat.w * clamp(lacc.w,0.0f,1.0f);\n", j, j);
}
else {
if( alpha.enablelighting )
if (alpha.enablelighting )
WRITE(p, "o.colors[%d] = mat * clamp(lacc,float4(0.0f,0.0f,0.0f,0.0f), float4(1.0f,1.0f,1.0f,1.0f));\n", j);
else WRITE(p, "o.colors[%d] = mat;\n", j);
}
@ -284,15 +284,15 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
}
// zero left over channels
for(int i = xfregs.nNumChans; i < 2; ++i) WRITE(p, "o.colors[%d] = 0;\n", i);
for (int i = xfregs.nNumChans; i < 2; ++i)
WRITE(p, "o.colors[%d] = 0;\n", i);
// transform texcoords
for(int i = 0; i < xfregs.numTexGens; ++i) {
for (int i = 0; i < xfregs.numTexGens; ++i) {
TexMtxInfo& texinfo = xfregs.texcoords[i].texmtxinfo;
WRITE(p, "{\n");
switch(texinfo.sourcerow) {
switch (texinfo.sourcerow) {
case XF_SRCGEOM_INROW:
_assert_( texinfo.inputform == XF_TEXINPUT_ABC1 );
WRITE(p, "float4 coord = rawpos;\n"); // pos.w is 1
@ -323,7 +323,7 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
break;
default:
_assert_(texinfo.sourcerow <= XF_SRCTEX7_INROW);
if ( components & (VertexLoader::VB_HAS_UV0<<(texinfo.sourcerow - XF_SRCTEX0_INROW)) )
if (components & (VertexLoader::VB_HAS_UV0<<(texinfo.sourcerow - XF_SRCTEX0_INROW)) )
WRITE(p, "float4 coord = float4(tex%d.x, tex%d.y, 1.0f, 1.0f);\n", texinfo.sourcerow - XF_SRCTEX0_INROW, texinfo.sourcerow - XF_SRCTEX0_INROW);
else
WRITE(p, "float4 coord = float4(0.0f, 0.0f, 1.0f, 1.0f);\n");
@ -331,9 +331,9 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
}
// firs transformation
switch(texinfo.texgentype) {
switch (texinfo.texgentype) {
case XF_TEXGEN_REGULAR:
if( components & (VertexLoader::VB_HAS_TEXMTXIDX0<<i) ) {
if (components & (VertexLoader::VB_HAS_TEXMTXIDX0<<i)) {
if (texinfo.projection == XF_TEXPROJ_STQ )
WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z+1].t), dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z+2].t));\n", i, i, i, i);
else {
@ -349,7 +349,7 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
break;
case XF_TEXGEN_EMBOSS_MAP: // calculate tex coords into bump map
if( components & (VertexLoader::VB_HAS_NRM1|VertexLoader::VB_HAS_NRM2) ) {
if (components & (VertexLoader::VB_HAS_NRM1|VertexLoader::VB_HAS_NRM2)) {
// transform the light dir into tangent space
WRITE(p, "ldir = normalize("I_LIGHTS".lights[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift);
WRITE(p, "o.tex%d.xyz = o.tex%d.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift);
@ -367,7 +367,7 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
break;
}
if(xfregs.bEnableDualTexTransform && texinfo.texgentype == XF_TEXGEN_REGULAR ) { // only works for regular tex gen types?
if(xfregs.bEnableDualTexTransform && texinfo.texgentype == XF_TEXGEN_REGULAR) { // only works for regular tex gen types?
if (xfregs.texcoords[i].postmtxinfo.normalize)
WRITE(p, "o.tex%d.xyz = normalize(o.tex%d.xyz);\n", i, i);
@ -383,11 +383,11 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
WRITE(p, "}\n");
}
if( ztexcoord >= 0 )
if (ztexcoord >= 0 )
WRITE(p, "o.tex%d.w = o.pos.z/o.pos.w;\n", ztexcoord);
// if( bpmem.fog.c_proj_fsel.fsel != 0 ) {
// switch(bpmem.fog.c_proj_fsel.fsel) {
// if (bpmem.fog.c_proj_fsel.fsel != 0) {
// switch (bpmem.fog.c_proj_fsel.fsel) {
// case 1: // linear
// break;
// case 4: // exp
@ -403,7 +403,7 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
// WRITE(p, "o.fog = o.pos.z/o.pos.w;\n");
// }
WRITE(p,"return o;\n}\n");
WRITE(p, "return o;\n}\n");
return text;
}
@ -412,12 +412,12 @@ char *GenerateVertexShader(u32 components, bool has_zbuffer_target)
char* GenerateLightShader(char* p, int index, const LitChannel& chan, const char* dest, int coloralpha)
{
const char* swizzle = "xyzw";
if( coloralpha == 1 ) swizzle = "xyz";
else if( coloralpha == 2 ) swizzle = "w";
if (coloralpha == 1 ) swizzle = "xyz";
else if (coloralpha == 2 ) swizzle = "w";
if( !(chan.attnfunc&1) ) {
if (!(chan.attnfunc&1)) {
// atten disabled
switch(chan.diffusefunc) {
switch (chan.diffusefunc) {
case LIGHTDIF_NONE:
WRITE(p, "%s.%s += "I_LIGHTS".lights[%d].col.%s;\n", dest, swizzle, index, swizzle);
break;
@ -433,20 +433,20 @@ char* GenerateLightShader(char* p, int index, const LitChannel& chan, const char
else { // spec and spot
WRITE(p, "ldir = "I_LIGHTS".lights[%d].pos.xyz - pos.xyz;\n", index);
if( chan.attnfunc == 3 ) { // spot
if (chan.attnfunc == 3) { // spot
WRITE(p, "dist2 = dot(ldir, ldir);\n"
"dist = sqrt(dist2);\n"
"ldir = ldir / dist;\n"
"attn = max(0.0f, dot(ldir, "I_LIGHTS".lights[%d].dir.xyz));\n",index);
WRITE(p, "attn = max(0.0f, dot("I_LIGHTS".lights[%d].cosatt.xyz, half3(1, attn, attn*attn))) / dot("I_LIGHTS".lights[%d].distatt.xyz, half3(1,dist,dist2));\n", index, index);
}
else if( chan.attnfunc == 1) { // specular
else if (chan.attnfunc == 1) { // specular
WRITE(p, "attn = dot(norm0, "I_LIGHTS".lights[%d].pos.xyz) > 0 ? max(0.0f, dot(norm0, "I_LIGHTS".lights[%d].dir.xyz)) : 0;\n", index, index);
WRITE(p, "ldir = half3(1,attn,attn*attn);\n");
WRITE(p, "attn = max(0.0f, dot("I_LIGHTS".lights[%d].cosatt.xyz, ldir)) / dot("I_LIGHTS".lights[%d].distatt.xyz, ldir);\n", index, index);
}
switch(chan.diffusefunc) {
switch (chan.diffusefunc) {
case LIGHTDIF_NONE:
WRITE(p, "%s.%s += attn * "I_LIGHTS".lights[%d].col.%s;\n", dest, swizzle, index, swizzle);
break;

15
Source/Plugins/Plugin_VideoOGL/Src/VertexShaderManager.cpp
View File

@ -19,10 +19,16 @@
#include "Globals.h"
#include "Profiler.h"
#include <Cg/cg.h>
#include <Cg/cgGL.h>
#include <math.h>
#include "Render.h"
#include "VertexShader.h"
#include "VertexShaderManager.h"
#include "VertexManager.h"
#include "VertexLoader.h"
#include "BPMemory.h"
#include "XFMemory.h"
@ -177,7 +183,7 @@ bool VertexShaderMngr::CompileVertexShader(VERTEXSHADER& vs, const char* pstrpro
char* pcompiledprog = (char*)cgGetProgramString(tempprog, CG_COMPILED_PROGRAM);
char* plocal = strstr(pcompiledprog, "program.local");
while( plocal != NULL ) {
while (plocal != NULL) {
const char* penv = " program.env";
memcpy(plocal, penv, 13);
plocal = strstr(plocal+13, "program.local");
@ -339,11 +345,8 @@ void VertexShaderMngr::SetConstants(VERTEXSHADER& vs)
2 * rawViewport[0], 2 * rawViewport[1],
(rawViewport[5] - rawViewport[2]) / 16777215.0f, rawViewport[5] / 16777215.0f);*/
// =======================================================================================
// Keep aspect ratio at 4:3
// -------------
// rawViewport[0] = 320, rawViewport[1] = -240
// -------------
int scissorXOff = bpmem.scissorOffset.x * 2 - 342;
int scissorYOff = bpmem.scissorOffset.y * 2 - 342;
float fourThree = 4.0f/3.0f;
@ -352,10 +355,10 @@ void VertexShaderMngr::SetConstants(VERTEXSHADER& vs)
int overfl; int xoffs = 0; int yoffs = 0;
int wid, hei, actualWid, actualHei;
int winw = nBackbufferWidth; int winh = nBackbufferHeight;
if(g_Config.bKeepAR)
if (g_Config.bKeepAR)
{
// Check if height or width is the limiting factor
if(ratio > 1) // then we are to wide and have to limit the width
if (ratio > 1) // then we are to wide and have to limit the width
{
wAdj = ratio;
hAdj = 1;

26
Source/Plugins/Plugin_VideoOGL/Src/main.cpp
View File

@ -27,6 +27,8 @@
#include "GUI/ConfigDlg.h"
#include "LookUpTables.h"
#include "ImageWrite.h"
#include "Render.h"
#include "GLInit.h"
#include "Fifo.h"
@ -34,6 +36,7 @@
#include "TextureMngr.h"
#include "BPStructs.h"
#include "VertexLoader.h"
#include "VertexManager.h"
#include "PixelShaderManager.h"
#include "VertexShaderManager.h"
#include "XFB.h"
@ -45,12 +48,9 @@
SVideoInitialize g_VideoInitialize;
#define VERSION_STRING "0.1"
// =======================================================================================
// Create debugging window. We can't use Show() here as usual because then DLL_PROCESS_DETACH will
// be called immediately. And if we use ShowModal() we block the main video window from appearing.
// So I've made a separate function called DoDllDebugger() that creates the window.
// -------------------
CDebugger* m_frame;
void DllDebugger(HWND _hParent)
{
@ -69,8 +69,6 @@ void DoDllDebugger()
m_frame = new CDebugger(NULL);
m_frame->Show();
}
// ===================
void GetDllInfo (PLUGIN_INFO* _PluginInfo)
{
@ -101,7 +99,7 @@ void DllConfig(HWND _hParent)
std::string resos[100];
int i = 0;
while(EnumDisplaySettings(NULL, iModeNum++, &dmi) != 0)
while (EnumDisplaySettings(NULL, iModeNum++, &dmi) != 0)
{
char szBuffer[100];
sprintf(szBuffer,"%dx%d", dmi.dmPelsWidth, dmi.dmPelsHeight);
@ -110,7 +108,7 @@ void DllConfig(HWND _hParent)
//create a check loop to check every pointer of resos to see if the res is added or not
int b = 0;
bool resFound = false;
while(b < i && !resFound)
while (b < i && !resFound)
{
//is the res already added?
resFound = (resos[b] == strBuffer);
@ -164,13 +162,10 @@ void DllConfig(HWND _hParent)
XFree(modes);
frame.ShowModal();
#else
//TODO
#endif
}
void Video_Initialize(SVideoInitialize* _pVideoInitialize)
{
if (_pVideoInitialize == NULL)
@ -197,7 +192,6 @@ void Video_Initialize(SVideoInitialize* _pVideoInitialize)
Renderer::AddMessage("Dolphin OpenGL Video Plugin v" VERSION_STRING ,5000);
}
void Video_DoState(unsigned char **ptr, int mode) {
// Clear all caches
@ -208,9 +202,7 @@ void Video_DoState(unsigned char **ptr, int mode) {
//PanicAlert("Saving/Loading state from OpenGL");
}
// =======================================================================================
// This is run after Video_Initialize() from the Core
// --------------
// This is called after Video_Initialize() from the Core
void Video_Prepare(void)
{
OpenGL_MakeCurrent();
@ -230,8 +222,6 @@ void Video_Prepare(void)
PixelShaderMngr::Init();
GL_REPORT_ERRORD();
}
// ==============
void Video_Shutdown(void)
{
@ -255,7 +245,6 @@ void Video_EnterLoop()
Fifo_EnterLoop(g_VideoInitialize);
}
void DebugLog(const char* _fmt, ...)
{
#if defined(_DEBUG) || defined(DEBUGFAST)
@ -271,7 +260,6 @@ void DebugLog(const char* _fmt, ...)
#endif
}
bool ScreenShot(TCHAR *File)
{
char str[64];
@ -292,7 +280,6 @@ bool ScreenShot(TCHAR *File)
return false;
}
BOOL Video_Screenshot(TCHAR* _szFilename)
{
if (ScreenShot(_szFilename))
@ -301,7 +288,6 @@ BOOL Video_Screenshot(TCHAR* _szFilename)
return FALSE;
}
void Video_UpdateXFB(u8* _pXFB, u32 _dwWidth, u32 _dwHeight, s32 _dwYOffset)
{
if(g_Config.bUseXFB)