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

Split a couple of huge files up, to make them easier to handle. 

git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@24 8ced0084-cf51-0410-be5f-012b33b47a6e
This commit is contained in:
hrydgard 2008-07-17 21:09:18 +00:00
parent af0e52554e
commit 91ccda69ef
18 changed files with 1696 additions and 1574 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Source/Plugins/Plugin_VideoDX9/Src/Utils.h
View File

@ -1,7 +1,6 @@
#ifndef _UTILS_H
#define _UTILS_H
#include "Common.h"
#include "main.h"
@ -17,7 +16,6 @@ extern float luts8tosfloat[256];
LRESULT CALLBACK AboutProc(HWND hDlg, UINT message, WPARAM wParam, LPARAM lParam);
void InitLUTs();
//#define RAM_MASK 0x1FFFFFF
inline u8 *Memory_GetPtr(u32 _uAddress)

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

@ -1301,6 +1301,14 @@
RelativePath=".\Src\PixelShader.h"
>
</File>
<File
RelativePath=".\Src\PixelShaderManager.cpp"
>
</File>
<File
RelativePath=".\Src\PixelShaderManager.h"
>
</File>
<File
RelativePath=".\Src\rasterfont.cpp"
>
@ -1349,6 +1357,14 @@
RelativePath=".\Src\VertexShader.h"
>
</File>
<File
RelativePath=".\Src\VertexShaderManager.cpp"
>
</File>
<File
RelativePath=".\Src\VertexShaderManager.h"
>
</File>
</Filter>
<File
RelativePath=".\Src\Fifo.cpp"

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

@ -23,8 +23,8 @@
#include "OpcodeDecoding.h"
#include "TextureMngr.h"
#include "TextureDecoder.h"
#include "VertexShader.h"
#include "PixelShader.h"
#include "VertexShaderManager.h"
#include "PixelShaderManager.h"
#define BPMEM_GENMODE 0x00
#define BPMEM_IND_MTX 0x06

1
Source/Plugins/Plugin_VideoOGL/Src/Globals.cpp
View File

@ -115,7 +115,6 @@ void Config::Save()
iniFile.Set("Hardware", "Fullscreen", bFullscreen);
iniFile.Set("Hardware", "RenderToMainframe", renderToMainframe);
iniFile.Set("Settings", "OverlayStats", bOverlayStats);
iniFile.Set("Settings", "OverlayStats", bOverlayStats);
iniFile.Set("Settings", "Postprocess", iPostprocessEffect);
iniFile.Set("Settings", "DLOptimize", iCompileDLsLevel);

2
Source/Plugins/Plugin_VideoOGL/Src/OpcodeDecoding.cpp
View File

@ -27,7 +27,7 @@
#include "Globals.h"
#include "OpcodeDecoding.h"
#include "VertexLoader.h"
#include "VertexShader.h"
#include "VertexShaderManager.h"
#include "TextureMngr.h"
#include "BPStructs.h"

580
Source/Plugins/Plugin_VideoOGL/Src/PixelShader.cpp
View File

@ -16,494 +16,12 @@
// http://code.google.com/p/dolphin-emu/
#include "Globals.h"
#include <math.h>
#include <cmath>
#include <assert.h>
#include "VertexShader.h"
#include "PixelShader.h"
#define I_COLORS "color"
#define I_KCOLORS "k"
#define I_ALPHA "alphaRef"
#define I_TEXDIMS "texdim"
#define I_ZBIAS "czbias"
#define I_INDTEXSCALE "cindscale"
#define I_INDTEXMTX "cindmtx"
#define C_COLORS 0
#define C_KCOLORS (C_COLORS+4)
#define C_ALPHA (C_KCOLORS+4)
#define C_TEXDIMS (C_ALPHA+1)
#define C_ZBIAS (C_TEXDIMS+8)
#define C_INDTEXSCALE (C_ZBIAS+2)
#define C_INDTEXMTX (C_INDTEXSCALE+2)
#define C_ENVCONST_END (C_INDTEXMTX+6)
#define C_COLORMATRIX (C_INDTEXMTX+6)
void WriteStage(char *&p, int n);
void WrapNonPow2Tex(char* &p, const char* var, int texmap);
void WriteAlphaCompare(char *&p, int num, int comp);
bool WriteAlphaTest(char *&p);
PixelShaderMngr::PSCache PixelShaderMngr::pshaders;
FRAGMENTSHADER* PixelShaderMngr::pShaderLast = NULL;
PixelShaderMngr::PIXELSHADERUID PixelShaderMngr::s_curuid;
static int s_nMaxPixelInstructions;
static int s_nColorsChanged[2]; // 0 - regular colors, 1 - k colors
static int s_nTexDimsChanged[2], s_nIndTexMtxChanged = 0; //min, max
static bool s_bAlphaChanged, s_bZBiasChanged, s_bIndTexScaleChanged;
static float lastRGBAfull[2][4][4] = {0};
static u32 lastAlpha = 0;
static u32 lastTexDims[8]={0};
static u32 lastZBias = 0;
// lower byte describes if a texture is nonpow2 or pow2
// next byte describes whether the repeat wrap mode is enabled for the s channel
// next byte is for t channel
static u32 s_texturemask = 0;
static int maptocoord[8]; // indexed by texture map, holds the texcoord associated with the map
static u32 maptocoord_mask=0;
static GLuint s_ColorMatrixProgram=0;
void PixelShaderMngr::Init()
{
s_nColorsChanged[0] = s_nColorsChanged[1] = 0;
s_nTexDimsChanged[0] = s_nTexDimsChanged[1] = -1;
s_nIndTexMtxChanged = 15;
s_bAlphaChanged = s_bZBiasChanged = s_bIndTexScaleChanged = true;
GL_REPORT_ERRORD();
for(int i = 0; i < 8; ++i) maptocoord[i] = -1;
maptocoord_mask = 0;
memset(lastRGBAfull, 0, sizeof(lastRGBAfull));
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB, &s_nMaxPixelInstructions);
int maxinst, maxattribs;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB, &maxinst);
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB, &maxattribs);
ERROR_LOG("pixel max_alu=%d, max_inst=%d, max_attrib=%d\n", s_nMaxPixelInstructions, maxinst, maxattribs);
char pmatrixprog[1024];
sprintf(pmatrixprog, "!!ARBfp1.0"
"TEMP R0;\n"
"TEMP R1;\n"
"TEX R0, fragment.texcoord[0], texture[0], RECT;\n"
"DP4 R1.w, R0, program.env[%d];\n"
"DP4 R1.z, R0, program.env[%d];\n"
"DP4 R1.x, R0, program.env[%d];\n"
"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);
GLenum err=GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
ERROR_LOG("Failed to create color matrix fragment program\n");
SAFE_RELEASE_PROG(s_ColorMatrixProgram);
}
}
void PixelShaderMngr::Shutdown()
{
SAFE_RELEASE_PROG(s_ColorMatrixProgram);
PSCache::iterator iter = pshaders.begin();
for (;iter!=pshaders.end();iter++)
iter->second.Destroy();
pshaders.clear();
}
FRAGMENTSHADER* PixelShaderMngr::GetShader()
{
DVSTARTPROFILE();
PIXELSHADERUID uid;
GetPixelShaderId(uid);
PSCache::iterator iter = pshaders.find(uid);
if (iter != pshaders.end()) {
iter->second.frameCount=frameCount;
PSCacheEntry &entry = iter->second;
if (&entry.shader != pShaderLast)
{
pShaderLast = &entry.shader;
}
return pShaderLast;
}
PSCacheEntry& newentry = pshaders[uid];
if (!GeneratePixelShader(newentry.shader)) {
ERROR_LOG("failed to create pixel shader\n");
return NULL;
}
//Make an entry in the table
newentry.frameCount=frameCount;
pShaderLast = &newentry.shader;
INCSTAT(stats.numPixelShadersCreated);
SETSTAT(stats.numPixelShadersAlive, pshaders.size());
return pShaderLast;
}
void PixelShaderMngr::Cleanup()
{
PSCache::iterator iter = pshaders.begin();
while(iter != pshaders.end()) {
PSCacheEntry &entry = iter->second;
if (entry.frameCount<frameCount-200) {
entry.Destroy();
#ifdef _WIN32
iter = pshaders.erase(iter);
#else
pshaders.erase(iter++); // (this is gcc standard!)
#endif
}
else
iter++;
}
SETSTAT(stats.numPixelShadersAlive,(int)pshaders.size());
}
bool PixelShaderMngr::CompilePixelShader(FRAGMENTSHADER& ps, const char* pstrprogram)
{
CGerror cgerr = cgGetError();
char stropt[64];
sprintf(stropt, "MaxLocalParams=32,NumInstructionSlots=%d", s_nMaxPixelInstructions);
#ifdef _WIN32
const char* opts[] = {"-profileopts",stropt,"-O2","-q",NULL};
#else
const char* opts[] = {"-profileopts",stropt,"-q",NULL};
#endif
CGprogram tempprog = cgCreateProgram(g_cgcontext, CG_SOURCE, pstrprogram, g_cgfProf, "main", opts);
if (!cgIsProgram(tempprog) || cgGetError() != CG_NO_ERROR) {
ERROR_LOG("Failed to create ps %s:\n", cgGetLastListing(g_cgcontext));
ERROR_LOG(pstrprogram);
return false;
}
char* pcompiledprog = (char*)cgGetProgramString(tempprog, CG_COMPILED_PROGRAM);
char* plocal = strstr(pcompiledprog, "program.local");
while( plocal != NULL ) {
const char* penv = " program.env";
memcpy(plocal, penv, 13);
plocal = strstr(plocal+13, "program.local");
}
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 ) {
poptions[0] = 'A';
poptions[1] = 'T';
poptions[2] = 'I';
}
}
//ERROR_LOG(pcompiledprog);
//ERROR_LOG(pstrprogram);
glGenProgramsARB( 1, &ps.glprogid );
glBindProgramARB( GL_FRAGMENT_PROGRAM_ARB, ps.glprogid );
glProgramStringARB( GL_FRAGMENT_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, (GLsizei)strlen(pcompiledprog), pcompiledprog);
GLenum err = GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
ERROR_LOG(pstrprogram);
ERROR_LOG(pcompiledprog);
}
cgDestroyProgram(tempprog);
#ifdef _DEBUG
ps.strprog = pstrprogram;
#endif
return true;
}
void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
{
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) ) {
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, baseind+j, &lastRGBAfull[i][j][0]);
}
}
s_nColorsChanged[i] = 0;
}
}
u32 newmask = 0;
for(u32 i = 0; i < 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;
SetTexDimsChanged(i);
}
}
if( maptocoord_mask != newmask ) {
u32 changes = maptocoord_mask ^ newmask;
for(int i = 0; i < 8; ++i) {
if( changes&(1<<i) ) {
SetTexDimsChanged(i);
}
if( !(newmask & (1<<i)) ) {
maptocoord[i] = -1;
}
}
maptocoord_mask = newmask;
}
if( s_nTexDimsChanged[0] >= 0 ) {
float fdims[4];
for(int i = s_nTexDimsChanged[0]; i <= s_nTexDimsChanged[1]; ++i) {
if( s_texturemask & (1<<i) ) {
if( maptocoord[i] >= 0 ) {
TCoordInfo& tc = bpmem.texcoords[maptocoord[i]];
fdims[0] = (float)(lastTexDims[i]&0xffff);
fdims[1] = (float)((lastTexDims[i]>>16)&0xfff);
fdims[2] = (float)(tc.s.scale_minus_1+1)/(float)(lastTexDims[i]&0xffff);
fdims[3] = (float)(tc.t.scale_minus_1+1)/(float)((lastTexDims[i]>>16)&0xfff);
}
else {
fdims[0] = (float)(lastTexDims[i]&0xffff);
fdims[1] = (float)((lastTexDims[i]>>16)&0xfff);
fdims[2] = 1.0f;
fdims[3] = 1.0f;
}
}
else {
if( maptocoord[i] >= 0 ) {
TCoordInfo& tc = bpmem.texcoords[maptocoord[i]];
fdims[0] = (float)(tc.s.scale_minus_1+1)/(float)(lastTexDims[i]&0xffff);
fdims[1] = (float)(tc.t.scale_minus_1+1)/(float)((lastTexDims[i]>>16)&0xfff);
fdims[2] = 1.0f/(float)(tc.s.scale_minus_1+1);
fdims[3] = 1.0f/(float)(tc.t.scale_minus_1+1);
}
else {
fdims[0] = 1.0f;
fdims[1] = 1.0f;
fdims[2] = 1.0f/(float)(lastTexDims[i]&0xffff);
fdims[3] = 1.0f/(float)((lastTexDims[i]>>16)&0xfff);
}
}
PRIM_LOG("texdims%d: %f %f %f %f\n", i, fdims[0], fdims[1], fdims[2], fdims[3]);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_TEXDIMS+i, fdims);
}
s_nTexDimsChanged[0] = s_nTexDimsChanged[1] = -1;
}
if( s_bAlphaChanged ) {
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_ALPHA, (lastAlpha&0xff)/255.0f, ((lastAlpha>>8)&0xff)/255.0f, 0, ((lastAlpha>>16)&0xff)/255.0f);
}
if( s_bZBiasChanged ) {
u32 bits;
float ffrac = 255.0f/256.0f;
float ftemp[4];
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;
break;
case 1:
bits = 16;
ftemp[0] = 0; ftemp[1] = ffrac/(256.0f*256.0f); ftemp[2] = ffrac/256.0f; ftemp[3] = ffrac;
break;
case 2:
bits = 24;
ftemp[0] = ffrac/(256.0f*256.0f); ftemp[1] = ffrac/256.0f; ftemp[2] = ffrac; ftemp[3] = 0;
break;
}
//ERROR_LOG("pixel=%x,%x, bias=%x\n", bpmem.zcontrol.pixel_format, bpmem.ztex2.type, lastZBias);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_ZBIAS, ftemp);
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_ZBIAS+1, 0, 0, 0, (float)( (((int)lastZBias<<8)>>8))/16777216.0f);
}
// indirect incoming texture scales, update all!
if( s_bIndTexScaleChanged ) {
float f[8];
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];
f[2*i] = bpmem.texscale[i/2].getScaleS(i&1) * (float)(tc.s.scale_minus_1+1) / (float)(lastTexDims[srctexmap]&0xffff);
f[2*i+1] = bpmem.texscale[i/2].getScaleT(i&1) * (float)(tc.t.scale_minus_1+1) / (float)((lastTexDims[srctexmap]>>16)&0xfff);
PRIM_LOG("tex indscale%d: %f %f\n", i, f[2*i], f[2*i+1]);
}
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_INDTEXSCALE, f);
if( bpmem.genMode.numindstages > 2 )
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, 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;
// xyz - static matrix
//TODO w - dynamic matrix scale / 256...... somehow / 4 works better
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_INDTEXMTX+2*i,
bpmem.indmtx[i].col0.ma * fscale, bpmem.indmtx[i].col1.mc * fscale, bpmem.indmtx[i].col2.me * fscale, fscale * 256.0f);
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_INDTEXMTX+2*i+1,
bpmem.indmtx[i].col0.mb * fscale, bpmem.indmtx[i].col1.md * fscale, bpmem.indmtx[i].col2.mf * fscale, fscale * 256.0f);
PRIM_LOG("indmtx%d: scale=%f, mat=(%f %f %f; %f %f %f)\n", i,
1024.0f*fscale, bpmem.indmtx[i].col0.ma * fscale, bpmem.indmtx[i].col1.mc * fscale, bpmem.indmtx[i].col2.me * fscale,
bpmem.indmtx[i].col0.mb * fscale, bpmem.indmtx[i].col1.md * fscale, bpmem.indmtx[i].col2.mf * fscale, fscale);
}
}
s_nIndTexMtxChanged = 0;
}
}
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;
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 ) {
lastAlpha = (lastAlpha&~0xffff)|(alpha.hex&0xffff);
s_bAlphaChanged = true;
}
}
void PixelShaderMngr::SetDestAlpha(const ConstantAlpha& alpha)
{
if( alpha.alpha != (lastAlpha>>16) ) {
lastAlpha = (lastAlpha&~0xff0000)|((alpha.hex&0xff)<<16);
s_bAlphaChanged = true;
}
}
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 ) {
lastTexDims[texmapid] = wh;
if( s_nTexDimsChanged[0] == -1 ) {
s_nTexDimsChanged[0] = s_nTexDimsChanged[1] = texmapid;
}
else {
if( s_nTexDimsChanged[0] > texmapid ) s_nTexDimsChanged[0] = texmapid;
else if( s_nTexDimsChanged[1] < texmapid ) s_nTexDimsChanged[1] = texmapid;
}
}
}
void PixelShaderMngr::SetZTetureBias(u32 bias)
{
if( lastZBias != bias ) {
s_bZBiasChanged = true;
lastZBias = bias;
}
}
void PixelShaderMngr::SetIndTexScaleChanged()
{
s_bIndTexScaleChanged = true;
}
void PixelShaderMngr::SetIndMatrixChanged(int matrixidx)
{
s_nIndTexMtxChanged |= 1 << matrixidx;
}
void PixelShaderMngr::SetGenModeChanged()
{
}
void PixelShaderMngr::SetTevCombinerChanged(int id)
{
}
void PixelShaderMngr::SetTevKSelChanged(int id)
{
}
void PixelShaderMngr::SetTevOrderChanged(int id)
{
}
void PixelShaderMngr::SetTevIndirectChanged(int id)
{
}
void PixelShaderMngr::SetZTetureOpChanged()
{
s_bZBiasChanged = true;
}
void PixelShaderMngr::SetTexturesUsed(u32 nonpow2tex)
{
if( s_texturemask != nonpow2tex ) {
u32 mask = s_texturemask ^ nonpow2tex;
for(int i = 0; i < 8; ++i) {
if( mask & (0x10101<<i) ) {
if( s_nTexDimsChanged[0] > i ) s_nTexDimsChanged[0] = i;
else if( s_nTexDimsChanged[1] < i ) s_nTexDimsChanged[1] = i;
}
}
s_texturemask = nonpow2tex;
}
}
void PixelShaderMngr::SetTexDimsChanged(int texmapid)
{
if( s_nTexDimsChanged[0] > texmapid ) s_nTexDimsChanged[0] = texmapid;
else if( s_nTexDimsChanged[1] < texmapid ) s_nTexDimsChanged[1] = texmapid;
SetIndTexScaleChanged();
}
void PixelShaderMngr::SetColorMatrix(const float* pmatrix, const float* pfConstAdd)
{
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX, pmatrix);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+1, pmatrix+4);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+2, pmatrix+8);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+3, pmatrix+12);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+4, pfConstAdd);
}
GLuint PixelShaderMngr::GetColorMatrixProgram()
{
return s_ColorMatrixProgram;
}
#include "VertexShader.h" // for texture projection mode
#include "PixelShaderManager.h"
// old tev->pixelshader notes
//
@ -513,6 +31,12 @@ GLuint PixelShaderMngr::GetColorMatrixProgram()
// according to GXTevColorArg table above
// output is given by .outreg
// tevtemp is set according to swapmodetables and
void WriteStage(char *&p, int n);
void WrapNonPow2Tex(char* &p, const char* var, int texmap);
void WriteAlphaCompare(char *&p, int num, int comp);
bool WriteAlphaTest(char *&p);
const float epsilon = 1.0f/255.0f;
const char *tevKSelTableC[] = // KCSEL
@ -719,90 +243,6 @@ const char* tevIndBiasAdd[] = {"-128.0f", "1.0f", "1.0f", "1.0f" }; // indexed b
const char* tevIndWrapStart[] = {"0", "256", "128", "64", "32", "16", "0.001" };
const char* tevIndFmtScale[] = {"255.0f", "31.0f", "15.0f", "8.0f" };
void PixelShaderMngr::GetPixelShaderId(PixelShaderMngr::PIXELSHADERUID& uid)
{
u32 projtexcoords = 0;
for (u32 i = 0; i < 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 )
projtexcoords |= 1<<texcoord;
}
}
u32 zbufrender = (Renderer::GetZBufferTarget()&&bpmem.zmode.updateenable)?1:0;
u32 zBufRenderToCol0 = Renderer::GetRenderMode()!=Renderer::RM_Normal;
uid.values[0] = (u32)bpmem.genMode.numtevstages|((u32)bpmem.genMode.numindstages<<4)|((u32)bpmem.genMode.numtexgens<<7)
|((u32)bpmem.dstalpha.enable<<11)|((u32)((bpmem.alphaFunc.hex>>16)&0xff)<<12)|(projtexcoords<<20)|((u32)bpmem.ztex2.op<<28)
|(zbufrender<<30)|(zBufRenderToCol0<<31);
s_curuid.values[0] = (s_curuid.values[0]&~0x0ff00000)|(projtexcoords<<20);
// swap table
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 < bpmem.genMode.numtevstages+1; ++i) {
TevStageCombiner::ColorCombiner &cc = bpmem.combiners[i].colorC;
TevStageCombiner::AlphaCombiner &ac = bpmem.combiners[i].alphaC;
u32 val0 = cc.hex&0xffffff;
u32 val1 = ac.hex&0xffffff;
val0 |= bpmem.tevksel[i/2].getKC(i&1)<<24;
val1 |= bpmem.tevksel[i/2].getKA(i&1)<<24;
pcurvalue[0] = val0;
pcurvalue[1] = val1;
pcurvalue+=2;
}
for(u32 i = 0; i < (bpmem.genMode.numtevstages+1)/2; ++i) {
u32 val0, val1;
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;
else val1 = (bpmem.tevorders[i].hex&0x380000)>>12;
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
u32 val0;
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) {
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 )
++pcurvalue;
uid.tevstages = (u32)(pcurvalue-&uid.values[0]-hdr);
for(u32 i = 0; i < bpmem.genMode.numindstages; ++i) {
u32 val = bpmem.tevind[i].hex&0x1fffff; // 21 bits
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;
}
}
uid.indstages = (u32)(pcurvalue-&uid.values[0]-2-uid.tevstages);
}
#define WRITE p+=sprintf
const char *swapColors = "rgba";
@ -822,7 +262,7 @@ void BuildSwapModeTable()
}
static char text[16384];
bool PixelShaderMngr::GeneratePixelShader(FRAGMENTSHADER& ps)
bool GeneratePixelShader(FRAGMENTSHADER& ps)
{
DVSTARTPROFILE();

139
Source/Plugins/Plugin_VideoOGL/Src/PixelShader.h
View File

@ -21,6 +21,25 @@
#include "Render.h"
#include "BPStructs.h"
#define I_COLORS "color"
#define I_KCOLORS "k"
#define I_ALPHA "alphaRef"
#define I_TEXDIMS "texdim"
#define I_ZBIAS "czbias"
#define I_INDTEXSCALE "cindscale"
#define I_INDTEXMTX "cindmtx"
#define C_COLORS 0
#define C_KCOLORS (C_COLORS+4)
#define C_ALPHA (C_KCOLORS+4)
#define C_TEXDIMS (C_ALPHA+1)
#define C_ZBIAS (C_TEXDIMS+8)
#define C_INDTEXSCALE (C_ZBIAS+2)
#define C_INDTEXMTX (C_INDTEXSCALE+2)
#define C_ENVCONST_END (C_INDTEXMTX+6)
#define C_COLORMATRIX (C_INDTEXMTX+6)
struct FRAGMENTSHADER
{
FRAGMENTSHADER() : glprogid(0) { }
@ -31,124 +50,6 @@ struct FRAGMENTSHADER
#endif
};
class PixelShaderMngr
{
class PIXELSHADERUID
{
public:
PIXELSHADERUID() { values = new u32[3+32+6+11]; tevstages = indstages = 0; }
~PIXELSHADERUID() { delete[] values; }
PIXELSHADERUID(const PIXELSHADERUID& r)
{
values = new u32[4+32+6+11];
tevstages = r.tevstages; indstages = r.indstages;
int N = tevstages + indstages + 3;
_assert_(N <= 4+32+6+11);
for(int i = 0; i < N; ++i)
values[i] = r.values[i];
}
bool operator <(const PIXELSHADERUID& _Right) const
{
if( values[0] < _Right.values[0] )
return true;
else if( values[0] > _Right.values[0] )
return false;
int N = tevstages + 3; // numTevStages*3/2+1
int i = 1;
for(; i < N; ++i) {
if( values[i] < _Right.values[i] )
return true;
else if( values[i] > _Right.values[i] )
return false;
}
N += indstages;
for(; i < N; ++i) {
if( values[i] < _Right.values[i] )
return true;
else if( values[i] > _Right.values[i] )
return false;
}
return false;
}
bool operator ==(const PIXELSHADERUID& _Right) const
{
if( values[0] != _Right.values[0] )
return false;
int N = tevstages + 3; // numTevStages*3/2+1
int i = 1;
for(; i < N; ++i) {
if( values[i] != _Right.values[i] )
return false;
}
N += indstages;
for(; i < N; ++i) {
if( values[i] != _Right.values[i] )
return false;
}
return true;
}
u32* values;
u16 tevstages, indstages;
};
struct PSCacheEntry
{
FRAGMENTSHADER shader;
int frameCount;
PSCacheEntry() : frameCount(0) {}
~PSCacheEntry() {}
void Destroy() {
glDeleteProgramsARB(1, &shader.glprogid);
}
};
typedef std::map<PIXELSHADERUID,PSCacheEntry> PSCache;
static FRAGMENTSHADER* pShaderLast; // last used shader
static PSCache pshaders;
static bool GeneratePixelShader(FRAGMENTSHADER& ps);
static void GetPixelShaderId(PIXELSHADERUID&);
static PIXELSHADERUID s_curuid; // the current pixel shader uid (progressively changed as memory is written)
public:
static void Init();
static void Cleanup();
static void Shutdown();
static FRAGMENTSHADER* GetShader();
static bool CompilePixelShader(FRAGMENTSHADER& ps, const char* pstrprogram);
static void SetConstants(FRAGMENTSHADER& ps); // sets pixel shader constants
// constant management, should be called after memory is committed
static void SetColorChanged(int type, int index);
static void SetAlpha(const AlphaFunc& alpha);
static void SetDestAlpha(const ConstantAlpha& alpha);
static void SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt);
static void SetZTetureBias(u32 bias);
static void SetIndTexScaleChanged();
static void SetIndMatrixChanged(int matrixidx);
static void SetGenModeChanged();
static void SetTevCombinerChanged(int id);
static void SetTevKSelChanged(int id);
static void SetTevOrderChanged(int id);
static void SetTevIndirectChanged(int id);
static void SetZTetureOpChanged();
static void SetTexturesUsed(u32 nonpow2tex);
static void SetTexDimsChanged(int texmapid);
static void SetColorMatrix(const float* pmatrix, const float* pfConstAdd);
static GLuint GetColorMatrixProgram();
};
bool GeneratePixelShader(FRAGMENTSHADER& ps);
#endif

567
Source/Plugins/Plugin_VideoOGL/Src/PixelShaderManager.cpp Normal file
View File

@ -0,0 +1,567 @@
// 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 <cmath>
#include "Common.h"
#include "VertexShader.h"
#include "PixelShaderManager.h"
#include "PixelShader.h"
PixelShaderMngr::PSCache PixelShaderMngr::pshaders;
FRAGMENTSHADER* PixelShaderMngr::pShaderLast = NULL;
PixelShaderMngr::PIXELSHADERUID PixelShaderMngr::s_curuid;
static int s_nMaxPixelInstructions;
static int s_nColorsChanged[2]; // 0 - regular colors, 1 - k colors
static int s_nTexDimsChanged[2], s_nIndTexMtxChanged = 0; //min, max
static bool s_bAlphaChanged, s_bZBiasChanged, s_bIndTexScaleChanged;
static float lastRGBAfull[2][4][4] = {0};
static u32 lastAlpha = 0;
static u32 lastTexDims[8]={0};
static u32 lastZBias = 0;
// lower byte describes if a texture is nonpow2 or pow2
// next byte describes whether the repeat wrap mode is enabled for the s channel
// next byte is for t channel
u32 s_texturemask = 0;
static int maptocoord[8]; // indexed by texture map, holds the texcoord associated with the map
static u32 maptocoord_mask=0;
static GLuint s_ColorMatrixProgram=0;
void PixelShaderMngr::Init()
{
s_nColorsChanged[0] = s_nColorsChanged[1] = 0;
s_nTexDimsChanged[0] = s_nTexDimsChanged[1] = -1;
s_nIndTexMtxChanged = 15;
s_bAlphaChanged = s_bZBiasChanged = s_bIndTexScaleChanged = true;
GL_REPORT_ERRORD();
for(int i = 0; i < 8; ++i) maptocoord[i] = -1;
maptocoord_mask = 0;
memset(lastRGBAfull, 0, sizeof(lastRGBAfull));
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB, &s_nMaxPixelInstructions);
int maxinst, maxattribs;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB, &maxinst);
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB, &maxattribs);
ERROR_LOG("pixel max_alu=%d, max_inst=%d, max_attrib=%d\n", s_nMaxPixelInstructions, maxinst, maxattribs);
char pmatrixprog[1024];
sprintf(pmatrixprog, "!!ARBfp1.0"
"TEMP R0;\n"
"TEMP R1;\n"
"TEX R0, fragment.texcoord[0], texture[0], RECT;\n"
"DP4 R1.w, R0, program.env[%d];\n"
"DP4 R1.z, R0, program.env[%d];\n"
"DP4 R1.x, R0, program.env[%d];\n"
"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);
GLenum err=GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
ERROR_LOG("Failed to create color matrix fragment program\n");
SAFE_RELEASE_PROG(s_ColorMatrixProgram);
}
}
void PixelShaderMngr::Shutdown()
{
SAFE_RELEASE_PROG(s_ColorMatrixProgram);
PSCache::iterator iter = pshaders.begin();
for (;iter!=pshaders.end();iter++)
iter->second.Destroy();
pshaders.clear();
}
FRAGMENTSHADER* PixelShaderMngr::GetShader()
{
DVSTARTPROFILE();
PIXELSHADERUID uid;
GetPixelShaderId(uid);
PSCache::iterator iter = pshaders.find(uid);
if (iter != pshaders.end()) {
iter->second.frameCount=frameCount;
PSCacheEntry &entry = iter->second;
if (&entry.shader != pShaderLast)
{
pShaderLast = &entry.shader;
}
return pShaderLast;
}
PSCacheEntry& newentry = pshaders[uid];
if (!GeneratePixelShader(newentry.shader)) {
ERROR_LOG("failed to create pixel shader\n");
return NULL;
}
//Make an entry in the table
newentry.frameCount=frameCount;
pShaderLast = &newentry.shader;
INCSTAT(stats.numPixelShadersCreated);
SETSTAT(stats.numPixelShadersAlive, pshaders.size());
return pShaderLast;
}
void PixelShaderMngr::Cleanup()
{
PSCache::iterator iter = pshaders.begin();
while(iter != pshaders.end()) {
PSCacheEntry &entry = iter->second;
if (entry.frameCount<frameCount-200) {
entry.Destroy();
#ifdef _WIN32
iter = pshaders.erase(iter);
#else
pshaders.erase(iter++); // (this is gcc standard!)
#endif
}
else
iter++;
}
SETSTAT(stats.numPixelShadersAlive,(int)pshaders.size());
}
bool PixelShaderMngr::CompilePixelShader(FRAGMENTSHADER& ps, const char* pstrprogram)
{
CGerror cgerr = cgGetError();
char stropt[64];
sprintf(stropt, "MaxLocalParams=32,NumInstructionSlots=%d", s_nMaxPixelInstructions);
#ifdef _WIN32
const char* opts[] = {"-profileopts",stropt,"-O2","-q",NULL};
#else
const char* opts[] = {"-profileopts",stropt,"-q",NULL};
#endif
CGprogram tempprog = cgCreateProgram(g_cgcontext, CG_SOURCE, pstrprogram, g_cgfProf, "main", opts);
if (!cgIsProgram(tempprog) || cgGetError() != CG_NO_ERROR) {
ERROR_LOG("Failed to create ps %s:\n", cgGetLastListing(g_cgcontext));
ERROR_LOG(pstrprogram);
return false;
}
char* pcompiledprog = (char*)cgGetProgramString(tempprog, CG_COMPILED_PROGRAM);
char* plocal = strstr(pcompiledprog, "program.local");
while( plocal != NULL ) {
const char* penv = " program.env";
memcpy(plocal, penv, 13);
plocal = strstr(plocal+13, "program.local");
}
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 ) {
poptions[0] = 'A';
poptions[1] = 'T';
poptions[2] = 'I';
}
}
//ERROR_LOG(pcompiledprog);
//ERROR_LOG(pstrprogram);
glGenProgramsARB( 1, &ps.glprogid );
glBindProgramARB( GL_FRAGMENT_PROGRAM_ARB, ps.glprogid );
glProgramStringARB( GL_FRAGMENT_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, (GLsizei)strlen(pcompiledprog), pcompiledprog);
GLenum err = GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
ERROR_LOG(pstrprogram);
ERROR_LOG(pcompiledprog);
}
cgDestroyProgram(tempprog);
#ifdef _DEBUG
ps.strprog = pstrprogram;
#endif
return true;
}
void PixelShaderMngr::SetConstants(FRAGMENTSHADER& ps)
{
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) ) {
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, baseind+j, &lastRGBAfull[i][j][0]);
}
}
s_nColorsChanged[i] = 0;
}
}
u32 newmask = 0;
for(u32 i = 0; i < 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;
SetTexDimsChanged(i);
}
}
if( maptocoord_mask != newmask ) {
u32 changes = maptocoord_mask ^ newmask;
for(int i = 0; i < 8; ++i) {
if( changes&(1<<i) ) {
SetTexDimsChanged(i);
}
if( !(newmask & (1<<i)) ) {
maptocoord[i] = -1;
}
}
maptocoord_mask = newmask;
}
if( s_nTexDimsChanged[0] >= 0 ) {
float fdims[4];
for(int i = s_nTexDimsChanged[0]; i <= s_nTexDimsChanged[1]; ++i) {
if( s_texturemask & (1<<i) ) {
if( maptocoord[i] >= 0 ) {
TCoordInfo& tc = bpmem.texcoords[maptocoord[i]];
fdims[0] = (float)(lastTexDims[i]&0xffff);
fdims[1] = (float)((lastTexDims[i]>>16)&0xfff);
fdims[2] = (float)(tc.s.scale_minus_1+1)/(float)(lastTexDims[i]&0xffff);
fdims[3] = (float)(tc.t.scale_minus_1+1)/(float)((lastTexDims[i]>>16)&0xfff);
}
else {
fdims[0] = (float)(lastTexDims[i]&0xffff);
fdims[1] = (float)((lastTexDims[i]>>16)&0xfff);
fdims[2] = 1.0f;
fdims[3] = 1.0f;
}
}
else {
if( maptocoord[i] >= 0 ) {
TCoordInfo& tc = bpmem.texcoords[maptocoord[i]];
fdims[0] = (float)(tc.s.scale_minus_1+1)/(float)(lastTexDims[i]&0xffff);
fdims[1] = (float)(tc.t.scale_minus_1+1)/(float)((lastTexDims[i]>>16)&0xfff);
fdims[2] = 1.0f/(float)(tc.s.scale_minus_1+1);
fdims[3] = 1.0f/(float)(tc.t.scale_minus_1+1);
}
else {
fdims[0] = 1.0f;
fdims[1] = 1.0f;
fdims[2] = 1.0f/(float)(lastTexDims[i]&0xffff);
fdims[3] = 1.0f/(float)((lastTexDims[i]>>16)&0xfff);
}
}
PRIM_LOG("texdims%d: %f %f %f %f\n", i, fdims[0], fdims[1], fdims[2], fdims[3]);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_TEXDIMS+i, fdims);
}
s_nTexDimsChanged[0] = s_nTexDimsChanged[1] = -1;
}
if( s_bAlphaChanged ) {
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_ALPHA, (lastAlpha&0xff)/255.0f, ((lastAlpha>>8)&0xff)/255.0f, 0, ((lastAlpha>>16)&0xff)/255.0f);
}
if( s_bZBiasChanged ) {
u32 bits;
float ffrac = 255.0f/256.0f;
float ftemp[4];
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;
break;
case 1:
bits = 16;
ftemp[0] = 0; ftemp[1] = ffrac/(256.0f*256.0f); ftemp[2] = ffrac/256.0f; ftemp[3] = ffrac;
break;
case 2:
bits = 24;
ftemp[0] = ffrac/(256.0f*256.0f); ftemp[1] = ffrac/256.0f; ftemp[2] = ffrac; ftemp[3] = 0;
break;
}
//ERROR_LOG("pixel=%x,%x, bias=%x\n", bpmem.zcontrol.pixel_format, bpmem.ztex2.type, lastZBias);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_ZBIAS, ftemp);
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_ZBIAS+1, 0, 0, 0, (float)( (((int)lastZBias<<8)>>8))/16777216.0f);
}
// indirect incoming texture scales, update all!
if( s_bIndTexScaleChanged ) {
float f[8];
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];
f[2*i] = bpmem.texscale[i/2].getScaleS(i&1) * (float)(tc.s.scale_minus_1+1) / (float)(lastTexDims[srctexmap]&0xffff);
f[2*i+1] = bpmem.texscale[i/2].getScaleT(i&1) * (float)(tc.t.scale_minus_1+1) / (float)((lastTexDims[srctexmap]>>16)&0xfff);
PRIM_LOG("tex indscale%d: %f %f\n", i, f[2*i], f[2*i+1]);
}
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_INDTEXSCALE, f);
if( bpmem.genMode.numindstages > 2 )
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, 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;
// xyz - static matrix
//TODO w - dynamic matrix scale / 256...... somehow / 4 works better
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_INDTEXMTX+2*i,
bpmem.indmtx[i].col0.ma * fscale, bpmem.indmtx[i].col1.mc * fscale, bpmem.indmtx[i].col2.me * fscale, fscale * 256.0f);
glProgramEnvParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, C_INDTEXMTX+2*i+1,
bpmem.indmtx[i].col0.mb * fscale, bpmem.indmtx[i].col1.md * fscale, bpmem.indmtx[i].col2.mf * fscale, fscale * 256.0f);
PRIM_LOG("indmtx%d: scale=%f, mat=(%f %f %f; %f %f %f)\n", i,
1024.0f*fscale, bpmem.indmtx[i].col0.ma * fscale, bpmem.indmtx[i].col1.mc * fscale, bpmem.indmtx[i].col2.me * fscale,
bpmem.indmtx[i].col0.mb * fscale, bpmem.indmtx[i].col1.md * fscale, bpmem.indmtx[i].col2.mf * fscale, fscale);
}
}
s_nIndTexMtxChanged = 0;
}
}
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;
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 ) {
lastAlpha = (lastAlpha&~0xffff)|(alpha.hex&0xffff);
s_bAlphaChanged = true;
}
}
void PixelShaderMngr::SetDestAlpha(const ConstantAlpha& alpha)
{
if( alpha.alpha != (lastAlpha>>16) ) {
lastAlpha = (lastAlpha&~0xff0000)|((alpha.hex&0xff)<<16);
s_bAlphaChanged = true;
}
}
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 ) {
lastTexDims[texmapid] = wh;
if( s_nTexDimsChanged[0] == -1 ) {
s_nTexDimsChanged[0] = s_nTexDimsChanged[1] = texmapid;
}
else {
if( s_nTexDimsChanged[0] > texmapid ) s_nTexDimsChanged[0] = texmapid;
else if( s_nTexDimsChanged[1] < texmapid ) s_nTexDimsChanged[1] = texmapid;
}
}
}
void PixelShaderMngr::SetZTetureBias(u32 bias)
{
if( lastZBias != bias ) {
s_bZBiasChanged = true;
lastZBias = bias;
}
}
void PixelShaderMngr::SetIndTexScaleChanged()
{
s_bIndTexScaleChanged = true;
}
void PixelShaderMngr::SetIndMatrixChanged(int matrixidx)
{
s_nIndTexMtxChanged |= 1 << matrixidx;
}
void PixelShaderMngr::SetGenModeChanged()
{
}
void PixelShaderMngr::SetTevCombinerChanged(int id)
{
}
void PixelShaderMngr::SetTevKSelChanged(int id)
{
}
void PixelShaderMngr::SetTevOrderChanged(int id)
{
}
void PixelShaderMngr::SetTevIndirectChanged(int id)
{
}
void PixelShaderMngr::SetZTetureOpChanged()
{
s_bZBiasChanged = true;
}
void PixelShaderMngr::SetTexturesUsed(u32 nonpow2tex)
{
if( s_texturemask != nonpow2tex ) {
u32 mask = s_texturemask ^ nonpow2tex;
for(int i = 0; i < 8; ++i) {
if( mask & (0x10101<<i) ) {
if( s_nTexDimsChanged[0] > i ) s_nTexDimsChanged[0] = i;
else if( s_nTexDimsChanged[1] < i ) s_nTexDimsChanged[1] = i;
}
}
s_texturemask = nonpow2tex;
}
}
void PixelShaderMngr::SetTexDimsChanged(int texmapid)
{
if( s_nTexDimsChanged[0] > texmapid ) s_nTexDimsChanged[0] = texmapid;
else if( s_nTexDimsChanged[1] < texmapid ) s_nTexDimsChanged[1] = texmapid;
SetIndTexScaleChanged();
}
void PixelShaderMngr::SetColorMatrix(const float* pmatrix, const float* pfConstAdd)
{
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX, pmatrix);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+1, pmatrix+4);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+2, pmatrix+8);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+3, pmatrix+12);
glProgramEnvParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, C_COLORMATRIX+4, pfConstAdd);
}
GLuint PixelShaderMngr::GetColorMatrixProgram()
{
return s_ColorMatrixProgram;
}
void PixelShaderMngr::GetPixelShaderId(PixelShaderMngr::PIXELSHADERUID& uid)
{
u32 projtexcoords = 0;
for (u32 i = 0; i < 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 )
projtexcoords |= 1<<texcoord;
}
}
u32 zbufrender = (Renderer::GetZBufferTarget()&&bpmem.zmode.updateenable)?1:0;
u32 zBufRenderToCol0 = Renderer::GetRenderMode()!=Renderer::RM_Normal;
uid.values[0] = (u32)bpmem.genMode.numtevstages|((u32)bpmem.genMode.numindstages<<4)|((u32)bpmem.genMode.numtexgens<<7)
|((u32)bpmem.dstalpha.enable<<11)|((u32)((bpmem.alphaFunc.hex>>16)&0xff)<<12)|(projtexcoords<<20)|((u32)bpmem.ztex2.op<<28)
|(zbufrender<<30)|(zBufRenderToCol0<<31);
s_curuid.values[0] = (s_curuid.values[0]&~0x0ff00000)|(projtexcoords<<20);
// swap table
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 < bpmem.genMode.numtevstages+1; ++i) {
TevStageCombiner::ColorCombiner &cc = bpmem.combiners[i].colorC;
TevStageCombiner::AlphaCombiner &ac = bpmem.combiners[i].alphaC;
u32 val0 = cc.hex&0xffffff;
u32 val1 = ac.hex&0xffffff;
val0 |= bpmem.tevksel[i/2].getKC(i&1)<<24;
val1 |= bpmem.tevksel[i/2].getKA(i&1)<<24;
pcurvalue[0] = val0;
pcurvalue[1] = val1;
pcurvalue+=2;
}
for(u32 i = 0; i < (bpmem.genMode.numtevstages+1)/2; ++i) {
u32 val0, val1;
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;
else val1 = (bpmem.tevorders[i].hex&0x380000)>>12;
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
u32 val0;
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) {
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 )
++pcurvalue;
uid.tevstages = (u32)(pcurvalue-&uid.values[0]-hdr);
for(u32 i = 0; i < bpmem.genMode.numindstages; ++i) {
u32 val = bpmem.tevind[i].hex&0x1fffff; // 21 bits
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;
}
}
uid.indstages = (u32)(pcurvalue-&uid.values[0]-2-uid.tevstages);
}

145
Source/Plugins/Plugin_VideoOGL/Src/PixelShaderManager.h Normal file
View File

@ -0,0 +1,145 @@
// 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 _PIXELSHADERMANAGER_H
#define _PIXELSHADERMANAGER_H
#include "PixelShader.h"
extern u32 s_texturemask;
class PixelShaderMngr
{
class PIXELSHADERUID
{
public:
PIXELSHADERUID() { values = new u32[3+32+6+11]; tevstages = indstages = 0; }
~PIXELSHADERUID() { delete[] values; }
PIXELSHADERUID(const PIXELSHADERUID& r)
{
values = new u32[4+32+6+11];
tevstages = r.tevstages; indstages = r.indstages;
int N = tevstages + indstages + 3;
_assert_(N <= 4+32+6+11);
for(int i = 0; i < N; ++i)
values[i] = r.values[i];
}
bool operator <(const PIXELSHADERUID& _Right) const
{
if( values[0] < _Right.values[0] )
return true;
else if( values[0] > _Right.values[0] )
return false;
int N = tevstages + 3; // numTevStages*3/2+1
int i = 1;
for(; i < N; ++i) {
if( values[i] < _Right.values[i] )
return true;
else if( values[i] > _Right.values[i] )
return false;
}
N += indstages;
for(; i < N; ++i) {
if( values[i] < _Right.values[i] )
return true;
else if( values[i] > _Right.values[i] )
return false;
}
return false;
}
bool operator ==(const PIXELSHADERUID& _Right) const
{
if( values[0] != _Right.values[0] )
return false;
int N = tevstages + 3; // numTevStages*3/2+1
int i = 1;
for(; i < N; ++i) {
if( values[i] != _Right.values[i] )
return false;
}
N += indstages;
for(; i < N; ++i) {
if( values[i] != _Right.values[i] )
return false;
}
return true;
}
u32* values;
u16 tevstages, indstages;
};
struct PSCacheEntry
{
FRAGMENTSHADER shader;
int frameCount;
PSCacheEntry() : frameCount(0) {}
~PSCacheEntry() {}
void Destroy() {
glDeleteProgramsARB(1, &shader.glprogid);
}
};
typedef std::map<PIXELSHADERUID,PSCacheEntry> PSCache;
static FRAGMENTSHADER* pShaderLast; // last used shader
static PSCache pshaders;
static void GetPixelShaderId(PIXELSHADERUID&);
static PIXELSHADERUID s_curuid; // the current pixel shader uid (progressively changed as memory is written)
public:
static void Init();
static void Cleanup();
static void Shutdown();
static FRAGMENTSHADER* GetShader();
static bool CompilePixelShader(FRAGMENTSHADER& ps, const char* pstrprogram);
static void SetConstants(FRAGMENTSHADER& ps); // sets pixel shader constants
// constant management, should be called after memory is committed
static void SetColorChanged(int type, int index);
static void SetAlpha(const AlphaFunc& alpha);
static void SetDestAlpha(const ConstantAlpha& alpha);
static void SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt);
static void SetZTetureBias(u32 bias);
static void SetIndTexScaleChanged();
static void SetIndMatrixChanged(int matrixidx);
static void SetGenModeChanged();
static void SetTevCombinerChanged(int id);
static void SetTevKSelChanged(int id);
static void SetTevOrderChanged(int id);
static void SetTevIndirectChanged(int id);
static void SetZTetureOpChanged();
static void SetTexturesUsed(u32 nonpow2tex);
static void SetTexDimsChanged(int texmapid);
static void SetColorMatrix(const float* pmatrix, const float* pfConstAdd);
static GLuint GetColorMatrixProgram();
};
#endif

2
Source/Plugins/Plugin_VideoOGL/Src/Render.cpp
View File

@ -25,7 +25,7 @@
#include "TextureMngr.h"
#include "rasterfont.h"
#include "VertexShader.h"
#include "PixelShader.h"
#include "PixelShaderManager.h"
#include "VertexLoader.h"

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

@ -27,8 +27,8 @@
#include "BPStructs.h"
#include "TextureDecoder.h"
#include "TextureMngr.h"
#include "PixelShader.h"
#include "VertexShader.h"
#include "PixelShaderManager.h"
#include "VertexShaderManager.h"
u8 *TextureMngr::temp = NULL;
TextureMngr::TexCache TextureMngr::textures;

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

@ -25,8 +25,8 @@
#include "BPStructs.h"
#include "DataReader.h"
#include "VertexShader.h"
#include "PixelShader.h"
#include "VertexShaderManager.h"
#include "PixelShaderManager.h"
#include "TextureMngr.h"
#include "MemoryUtil.h"

782
Source/Plugins/Plugin_VideoOGL/Src/VertexShader.cpp
View File

@ -20,790 +20,18 @@
#include <math.h>
#include "Render.h"
#include "VertexShader.h"
#include "VertexShaderManager.h"
#include "BPStructs.h"
#include "VertexLoader.h"
// shader variables
#define I_POSNORMALMATRIX "cpnmtx"
#define I_PROJECTION "cproj"
#define I_MATERIALS "cmtrl"
#define I_LIGHTS "clights"
#define I_TEXMATRICES "ctexmtx"
#define I_TRANSFORMMATRICES "ctrmtx"
#define I_NORMALMATRICES "cnmtx"
#define I_POSTTRANSFORMMATRICES "cpostmtx"
#define I_FOGPARAMS "cfog"
#define C_POSNORMALMATRIX 0
#define C_PROJECTION (C_POSNORMALMATRIX+6)
#define C_MATERIALS (C_PROJECTION+4)
#define C_LIGHTS (C_MATERIALS+4)
#define C_TEXMATRICES (C_LIGHTS+40)
#define C_TRANSFORMMATRICES (C_TEXMATRICES+24)
#define C_NORMALMATRICES (C_TRANSFORMMATRICES+64)
#define C_POSTTRANSFORMMATRICES (C_NORMALMATRICES+32)
#define C_FOGPARAMS (C_POSTTRANSFORMMATRICES+64)
VertexShaderMngr::VSCache VertexShaderMngr::vshaders;
VERTEXSHADER* VertexShaderMngr::pShaderLast = NULL;
TMatrixIndexA VertexShaderMngr::MatrixIndexA;
TMatrixIndexB VertexShaderMngr::MatrixIndexB;
float VertexShaderMngr::rawViewport[6] = {0};
float VertexShaderMngr::rawProjection[7] = {0};
float GC_ALIGNED16(g_fProjectionMatrix[16]);
static int s_nMaxVertexInstructions;
////////////////////////
// Internal Variables //
////////////////////////
XFRegisters xfregs;
static u32 xfmem[XFMEM_SIZE];
static float s_fMaterials[16];
// track changes
static bool bTexMatricesChanged[2], bPosNormalMatrixChanged, bProjectionChanged, bViewportChanged;
int nMaterialsChanged;
static int nTransformMatricesChanged[2]; // min,max
static int nNormalMatricesChanged[2]; // min,max
static int nPostTransformMatricesChanged[2]; // min,max
static int nLightsChanged[2]; // min,max
void VertexShaderMngr::Init()
{
nTransformMatricesChanged[0] = nTransformMatricesChanged[1] = -1;
nNormalMatricesChanged[0] = nNormalMatricesChanged[1] = -1;
nPostTransformMatricesChanged[0] = nPostTransformMatricesChanged[1] = -1;
nLightsChanged[0] = nLightsChanged[1] = -1;
bTexMatricesChanged[0] = bTexMatricesChanged[1] = false;
bPosNormalMatrixChanged = bProjectionChanged = bViewportChanged = false;
nMaterialsChanged = 0;
memset(&xfregs, 0, sizeof(xfregs));
memset(xfmem, 0, sizeof(xfmem));
glGetProgramivARB(GL_VERTEX_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB, &s_nMaxVertexInstructions);
}
void VertexShaderMngr::Shutdown()
{
VSCache::iterator iter = vshaders.begin();
for (;iter!=vshaders.end();iter++)
iter->second.Destroy();
vshaders.clear();
}
VERTEXSHADER* VertexShaderMngr::GetShader(u32 components)
{
DVSTARTPROFILE();
VERTEXSHADERUID uid;
GetVertexShaderId(uid, components);
VSCache::iterator iter = vshaders.find(uid);
if (iter != vshaders.end()) {
iter->second.frameCount=frameCount;
VSCacheEntry &entry = iter->second;
if (&entry.shader != pShaderLast) {
pShaderLast = &entry.shader;
}
return pShaderLast;
}
VSCacheEntry& entry = vshaders[uid];
if (!GenerateVertexShader(entry.shader, components)) {
ERROR_LOG("failed to create vertex shader\n");
}
//Make an entry in the table
entry.frameCount=frameCount;
pShaderLast = &entry.shader;
INCSTAT(stats.numVertexShadersCreated);
SETSTAT(stats.numVertexShadersAlive,vshaders.size());
return pShaderLast;
}
void VertexShaderMngr::Cleanup()
{
VSCache::iterator iter=vshaders.begin();
while (iter != vshaders.end()) {
VSCacheEntry &entry = iter->second;
if (entry.frameCount < frameCount-200) {
entry.Destroy();
#ifdef _WIN32
iter = vshaders.erase(iter);
#else
vshaders.erase(iter++);
#endif
}
else {
++iter;
}
}
// static int frame = 0;
// if( frame++ > 30 ) {
// VSCache::iterator iter=vshaders.begin();
// while(iter!=vshaders.end()) {
// iter->second.Destroy();
// ++iter;
// }
// vshaders.clear();
// }
SETSTAT(stats.numPixelShadersAlive,vshaders.size());
}
bool VertexShaderMngr::CompileVertexShader(VERTEXSHADER& vs, const char* pstrprogram)
{
char stropt[64];
sprintf(stropt, "MaxLocalParams=256,MaxInstructions=%d", s_nMaxVertexInstructions);
#ifdef _WIN32
const char* opts[] = {"-profileopts",stropt,"-O2", "-q", NULL};
#else
const char* opts[] = {"-profileopts",stropt,"-q", NULL};
#endif
CGprogram tempprog = cgCreateProgram(g_cgcontext, CG_SOURCE, pstrprogram, g_cgvProf, "main", opts);
if (!cgIsProgram(tempprog) || cgGetError() != CG_NO_ERROR) {
ERROR_LOG("Failed to load vs %s:\n", cgGetLastListing(g_cgcontext));
ERROR_LOG(pstrprogram);
return false;
}
//ERROR_LOG(pstrprogram);
//ERROR_LOG("id: %d\n", g_Config.iSaveTargetId);
char* pcompiledprog = (char*)cgGetProgramString(tempprog, CG_COMPILED_PROGRAM);
char* plocal = strstr(pcompiledprog, "program.local");
while( plocal != NULL ) {
const char* penv = " program.env";
memcpy(plocal, penv, 13);
plocal = strstr(plocal+13, "program.local");
}
glGenProgramsARB( 1, &vs.glprogid );
glBindProgramARB( GL_VERTEX_PROGRAM_ARB, vs.glprogid );
glProgramStringARB( GL_VERTEX_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, (GLsizei)strlen(pcompiledprog), pcompiledprog);
GLenum err = GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
ERROR_LOG(pstrprogram);
ERROR_LOG(pcompiledprog);
}
cgDestroyProgram(tempprog);
#ifdef _DEBUG
vs.strprog = pstrprogram;
#endif
return true;
}
// TODO: this array is misdeclared. Why teh f** does it go through the compilers?
const u16 s_mtrltable[16][2] = {0, 0, 0, 1, 1, 1, 0, 2,
2, 1, 0, 3, 1, 2, 0, 3,
3, 1, 0, 4, 1, 3, 0, 4,
2, 2, 0, 4, 1, 3, 0, 4};
/// syncs the shader constant buffers with xfmem
void VertexShaderMngr::SetConstants(VERTEXSHADER& vs)
{
//nTransformMatricesChanged[0] = 0; nTransformMatricesChanged[1] = 256;
//nNormalMatricesChanged[0] = 0; nNormalMatricesChanged[1] = 96;
//nPostTransformMatricesChanged[0] = 0; nPostTransformMatricesChanged[1] = 256;
//nLightsChanged[0] = 0; nLightsChanged[1] = 0x80;
//bPosNormalMatrixChanged = true;
//bTexMatricesChanged[0] = bTexMatricesChanged[1] = true;
//bProjectionChanged = true;
// bPosNormalMatrixChanged = bTexMatricesChanged[0] = bTexMatricesChanged[1] = true;
// nMaterialsChanged = 15;
if (nTransformMatricesChanged[0] >= 0) {
int startn = nTransformMatricesChanged[0]/4;
int endn = (nTransformMatricesChanged[1]+3)/4;
const float* pstart = (const float*)&xfmem[startn*4];
for(int i = startn; i < endn; ++i, pstart += 4)
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TRANSFORMMATRICES+i, pstart);
nTransformMatricesChanged[0] = nTransformMatricesChanged[1] = -1;
}
if (nNormalMatricesChanged[0] >= 0) {
int startn = nNormalMatricesChanged[0]/3;
int endn = (nNormalMatricesChanged[1]+2)/3;
const float* pnstart = (const float*)&xfmem[XFMEM_NORMALMATRICES+3*startn];
for(int i = startn; i < endn; ++i, pnstart += 3)
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_NORMALMATRICES+i, pnstart);
nNormalMatricesChanged[0] = nNormalMatricesChanged[1] = -1;
}
if (nPostTransformMatricesChanged[0] >= 0) {
int startn = nPostTransformMatricesChanged[0]/4;
int endn = (nPostTransformMatricesChanged[1]+3)/4;
const float* pstart = (const float*)&xfmem[XFMEM_POSTMATRICES+startn*4];
for(int i = startn; i < endn; ++i, pstart += 4)
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSTTRANSFORMMATRICES+i, pstart);
}
if (nLightsChanged[0] >= 0) {
// lights don't have a 1 to 1 mapping, the color component needs to be converted to 4 floats
int istart = nLightsChanged[0]/0x10;
int iend = (nLightsChanged[1]+15)/0x10;
const float* xfmemptr = (const float*)&xfmem[0x10*istart+XFMEM_LIGHTS];
for(int i = istart; i < iend; ++i) {
u32 color = *(const u32*)(xfmemptr+3);
glProgramEnvParameter4fARB(GL_VERTEX_PROGRAM_ARB, C_LIGHTS+5*i,
((color>>24)&0xFF)/255.0f, ((color>>16)&0xFF)/255.0f, ((color>>8)&0xFF)/255.0f, ((color)&0xFF)/255.0f);
xfmemptr += 4;
for(int j = 0; j < 4; ++j, xfmemptr += 3) {
if( j == 1 && fabs(xfmemptr[0]) < 0.00001f && fabs(xfmemptr[1]) < 0.00001f && fabs(xfmemptr[2]) < 0.00001f) {
// dist atten, make sure not equal to 0!!!
glProgramEnvParameter4fARB(GL_VERTEX_PROGRAM_ARB, C_LIGHTS+5*i+j+1, 0.00001f, xfmemptr[1], xfmemptr[2], 0);
}
else
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_LIGHTS+5*i+j+1, xfmemptr);
}
}
nLightsChanged[0] = nLightsChanged[1] = -1;
}
if (nMaterialsChanged) {
for(int i = 0; i < 4; ++i) {
if( nMaterialsChanged&(1<<i) )
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_MATERIALS+i, &s_fMaterials[4*i]);
}
nMaterialsChanged = 0;
}
if (bPosNormalMatrixChanged) {
bPosNormalMatrixChanged = false;
float* pos = (float*)xfmem + MatrixIndexA.PosNormalMtxIdx * 4;
float* norm = (float*)xfmem + XFMEM_NORMALMATRICES + 3 * (MatrixIndexA.PosNormalMtxIdx & 31);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX, pos);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+1, pos+4);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+2, pos+8);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+3, norm);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+4, norm+3);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+5, norm+6);
}
if (bTexMatricesChanged[0]) {
bTexMatricesChanged[0] = false;
float* fptrs[] = {(float*)xfmem + MatrixIndexA.Tex0MtxIdx * 4, (float*)xfmem + MatrixIndexA.Tex1MtxIdx * 4,
(float*)xfmem + MatrixIndexA.Tex2MtxIdx * 4, (float*)xfmem + MatrixIndexA.Tex3MtxIdx * 4 };
for(int i = 0; i < 4; ++i) {
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i, fptrs[i]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+1, fptrs[i]+4);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+2, fptrs[i]+8);
}
}
if (bTexMatricesChanged[1]) {
bTexMatricesChanged[1] = false;
float* fptrs[] = {(float*)xfmem + MatrixIndexB.Tex4MtxIdx * 4, (float*)xfmem + MatrixIndexB.Tex5MtxIdx * 4,
(float*)xfmem + MatrixIndexB.Tex6MtxIdx * 4, (float*)xfmem + MatrixIndexB.Tex7MtxIdx * 4 };
for(int i = 0; i < 4; ++i) {
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+12, fptrs[i]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+12+1, fptrs[i]+4);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+12+2, fptrs[i]+8);
}
}
if (bViewportChanged) {
bViewportChanged = false;
// reversed gxsetviewport(xorig, yorig, width, height, nearz, farz)
// [0] = width/2
// [1] = height/2
// [2] = 16777215 * (farz-nearz)
// [3] = xorig + width/2 + 342
// [4] = yorig + height/2 + 342
// [5] = 16777215 * farz
INFO_LOG("view: topleft=(%f,%f), wh=(%f,%f), z=(%f,%f)\n",rawViewport[3]-rawViewport[0]-342,rawViewport[4]+rawViewport[1]-342,
2 * rawViewport[0], 2 * rawViewport[1], (rawViewport[5]-rawViewport[2])/16777215.0f, rawViewport[5]/16777215.0f);
glViewport((int)(rawViewport[3]-rawViewport[0]-342)<<g_AAx,Renderer::GetTargetHeight()-((int)(rawViewport[4]-rawViewport[1]-342)<<g_AAy), abs((int)(2 * rawViewport[0])), abs((int)(2 * rawViewport[1])));
glDepthRange((rawViewport[5]-rawViewport[2])/16777215.0f, rawViewport[5]/16777215.0f);
}
if (bProjectionChanged) {
bProjectionChanged = false;
if (rawProjection[6] == 0) {
g_fProjectionMatrix[0] = rawProjection[0];
g_fProjectionMatrix[1] = 0.0f;
g_fProjectionMatrix[2] = rawProjection[1];
g_fProjectionMatrix[3] = 0;//-0.5f/Renderer::GetTargetWidth();
g_fProjectionMatrix[4] = 0.0f;
g_fProjectionMatrix[5] = rawProjection[2];
g_fProjectionMatrix[6] = rawProjection[3];
g_fProjectionMatrix[7] = 0;//+0.5f/Renderer::GetTargetHeight();
g_fProjectionMatrix[8] = 0.0f;
g_fProjectionMatrix[9] = 0.0f;
g_fProjectionMatrix[10] = -(1-rawProjection[4]);
g_fProjectionMatrix[11] = rawProjection[5];
g_fProjectionMatrix[12] = 0.0f;
g_fProjectionMatrix[13] = 0.0f;
g_fProjectionMatrix[14] = -1.0f;
g_fProjectionMatrix[15] = 0.0f;
}
else {
g_fProjectionMatrix[0] = rawProjection[0];
g_fProjectionMatrix[1] = 0.0f;
g_fProjectionMatrix[2] = 0.0f;
g_fProjectionMatrix[3] = rawProjection[1];
g_fProjectionMatrix[4] = 0.0f;
g_fProjectionMatrix[5] = rawProjection[2];
g_fProjectionMatrix[6] = 0.0f;
g_fProjectionMatrix[7] = rawProjection[3];
g_fProjectionMatrix[8] = 0.0f;
g_fProjectionMatrix[9] = 0.0f;
g_fProjectionMatrix[10] = rawProjection[4];
g_fProjectionMatrix[11] = -(-1 - rawProjection[5]);
g_fProjectionMatrix[12] = 0;
g_fProjectionMatrix[13] = 0;
g_fProjectionMatrix[14] = 0.0f;
g_fProjectionMatrix[15] = 1.0f;
}
PRIM_LOG("Projection: %f %f %f %f %f %f\n",rawProjection[0], rawProjection[1], rawProjection[2], rawProjection[3], rawProjection[4], rawProjection[5]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION, &g_fProjectionMatrix[0]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION+1, &g_fProjectionMatrix[4]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION+2, &g_fProjectionMatrix[8]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION+3, &g_fProjectionMatrix[12]);
}
}
void VertexShaderMngr::InvalidateXFRange(int start, int end)
{
if( ((u32)start >= MatrixIndexA.PosNormalMtxIdx*4 && (u32)start < MatrixIndexA.PosNormalMtxIdx*4+12) ||
((u32)start >= XFMEM_NORMALMATRICES+(MatrixIndexA.PosNormalMtxIdx&31)*3 && (u32)start < XFMEM_NORMALMATRICES+(MatrixIndexA.PosNormalMtxIdx&31)*3+9) ) {
bPosNormalMatrixChanged = true;
}
if (((u32)start >= MatrixIndexA.Tex0MtxIdx*4 && (u32)start < MatrixIndexA.Tex0MtxIdx*4+12) ||
((u32)start >= MatrixIndexA.Tex1MtxIdx*4 && (u32)start < MatrixIndexA.Tex1MtxIdx*4+12) ||
((u32)start >= MatrixIndexA.Tex2MtxIdx*4 && (u32)start < MatrixIndexA.Tex2MtxIdx*4+12) ||
((u32)start >= MatrixIndexA.Tex3MtxIdx*4 && (u32)start < MatrixIndexA.Tex3MtxIdx*4+12)) {
bTexMatricesChanged[0] = true;
}
if (((u32)start >= MatrixIndexB.Tex4MtxIdx*4 && (u32)start < MatrixIndexB.Tex4MtxIdx*4+12) ||
((u32)start >= MatrixIndexB.Tex5MtxIdx*4 && (u32)start < MatrixIndexB.Tex5MtxIdx*4+12) ||
((u32)start >= MatrixIndexB.Tex6MtxIdx*4 && (u32)start < MatrixIndexB.Tex6MtxIdx*4+12) ||
((u32)start >= MatrixIndexB.Tex7MtxIdx*4 && (u32)start < MatrixIndexB.Tex7MtxIdx*4+12)) {
bTexMatricesChanged[1] = true;
}
if (start < XFMEM_POSMATRICES_END ) {
if (nTransformMatricesChanged[0] == -1) {
nTransformMatricesChanged[0] = start;
nTransformMatricesChanged[1] = end>XFMEM_POSMATRICES_END?XFMEM_POSMATRICES_END:end;
}
else {
if (nTransformMatricesChanged[0] > start) nTransformMatricesChanged[0] = start;
if (nTransformMatricesChanged[1] < end) nTransformMatricesChanged[1] = end>XFMEM_POSMATRICES_END?XFMEM_POSMATRICES_END:end;
}
}
if (start < XFMEM_NORMALMATRICES_END && end > XFMEM_NORMALMATRICES ) {
int _start = start < XFMEM_NORMALMATRICES ? 0 : start-XFMEM_NORMALMATRICES;
int _end = end < XFMEM_NORMALMATRICES_END ? end-XFMEM_NORMALMATRICES : XFMEM_NORMALMATRICES_END-XFMEM_NORMALMATRICES;
if (nNormalMatricesChanged[0] == -1 ) {
nNormalMatricesChanged[0] = _start;
nNormalMatricesChanged[1] = _end;
}
else {
if (nNormalMatricesChanged[0] > _start) nNormalMatricesChanged[0] = _start;
if (nNormalMatricesChanged[1] < _end) nNormalMatricesChanged[1] = _end;
}
}
if (start < XFMEM_POSTMATRICES_END && end > XFMEM_POSTMATRICES ) {
int _start = start < XFMEM_POSTMATRICES ? XFMEM_POSTMATRICES : start-XFMEM_POSTMATRICES;
int _end = end < XFMEM_POSTMATRICES_END ? end-XFMEM_POSTMATRICES : XFMEM_POSTMATRICES_END-XFMEM_POSTMATRICES;
if (nPostTransformMatricesChanged[0] == -1 ) {
nPostTransformMatricesChanged[0] = _start;
nPostTransformMatricesChanged[1] = _end;
}
else {
if (nPostTransformMatricesChanged[0] > _start) nPostTransformMatricesChanged[0] = _start;
if (nPostTransformMatricesChanged[1] < _end) nPostTransformMatricesChanged[1] = _end;
}
}
if (start < XFMEM_LIGHTS_END && end > XFMEM_LIGHTS) {
int _start = start < XFMEM_LIGHTS ? XFMEM_LIGHTS : start-XFMEM_LIGHTS;
int _end = end < XFMEM_LIGHTS_END ? end-XFMEM_LIGHTS : XFMEM_LIGHTS_END-XFMEM_LIGHTS;
if (nLightsChanged[0] == -1 ) {
nLightsChanged[0] = _start;
nLightsChanged[1] = _end;
}
else {
if (nLightsChanged[0] > _start) nLightsChanged[0] = _start;
if (nLightsChanged[1] < _end) nLightsChanged[1] = _end;
}
}
}
void VertexShaderMngr::SetTexMatrixChangedA(u32 Value)
{
if (MatrixIndexA.Hex != Value) {
VertexManager::Flush();
if (MatrixIndexA.PosNormalMtxIdx != (Value&0x3f))
bPosNormalMatrixChanged = true;
bTexMatricesChanged[0] = true;
MatrixIndexA.Hex = Value;
}
}
void VertexShaderMngr::SetTexMatrixChangedB(u32 Value)
{
if (MatrixIndexB.Hex != Value) {
VertexManager::Flush();
bTexMatricesChanged[1] = true;
MatrixIndexB.Hex = Value;
}
}
void VertexShaderMngr::SetViewport(float* _Viewport)
{
// check for paper mario
for (size_t i = 0; i < ARRAYSIZE(rawViewport); ++i) {
if( *(u32*)(_Viewport + i) == 0x7f800000 )
return; // invalid number
}
memcpy(rawViewport, _Viewport, sizeof(rawViewport));
bViewportChanged = true;
}
void VertexShaderMngr::SetViewportChanged()
{
bViewportChanged = true;
}
void VertexShaderMngr::SetProjection(float* _pProjection, int constantIndex)
{
memcpy(rawProjection, _pProjection, sizeof(rawProjection));
bProjectionChanged = true;
}
size_t VertexShaderMngr::SaveLoadState(char *ptr, BOOL save)
{
BEGINSAVELOAD;
SAVELOAD(&xfregs,sizeof(xfregs));
SAVELOAD(xfmem,XFMEM_SIZE*sizeof(u32));
SAVELOAD(rawViewport,sizeof(rawViewport));
SAVELOAD(rawProjection,sizeof(rawProjection));
SAVELOAD(&MatrixIndexA,sizeof(TMatrixIndexA));
SAVELOAD(&MatrixIndexB,sizeof(TMatrixIndexB));
if (!save) {
// invalidate all
InvalidateXFRange(0,0x1000);
}
ENDSAVELOAD;
}
// LoadXFReg 0x10
void VertexShaderMngr::LoadXFReg(u32 transferSize, u32 baseAddress, u32 *pData)
{
u32 address = baseAddress;
for (int i=0; i<(int)transferSize; i++)
{
address = baseAddress + i;
// Setup a Matrix
if (address < 0x1000)
{
VertexManager::Flush();
InvalidateXFRange(address, address+transferSize);
//PRIM_LOG("xfmem write: 0x%x-0x%x\n", address, address+transferSize);
u32* p1 = &xfmem[address];
memcpy_gc(p1, &pData[i], transferSize*4);
i += transferSize;
}
else if (address<0x2000)
{
u32 data = pData[i];
switch (address)
{
case 0x1000: // error
break;
case 0x1001: // diagnostics
break;
case 0x1002: // internal state 0
break;
case 0x1003: // internal state 1
break;
case 0x1004: // xf_clock
break;
case 0x1005: // clipdisable
if (data & 1) { // disable clipping detection
}
if (data & 2) { // disable trivial rejection
}
if (data & 4) { // disable cpoly clipping acceleration
}
break;
case 0x1006: //SetGPMetric
break;
case 0x1008: //__GXXfVtxSpecs, wrote 0004
xfregs.hostinfo = *(INVTXSPEC*)&data;
break;
case 0x1009: //GXSetNumChans (no)
if (xfregs.nNumChans != (data&3) ) {
VertexManager::Flush();
xfregs.nNumChans = data&3;
}
break;
case 0x100a: //GXSetChanAmbientcolor
if (xfregs.colChans[0].ambColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 1;
xfregs.colChans[0].ambColor = data;
s_fMaterials[0] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[1] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[2] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[3] = ((data)&0xFF)/255.0f;
}
break;
case 0x100b: //GXSetChanAmbientcolor
if (xfregs.colChans[1].ambColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 2;
xfregs.colChans[1].ambColor = data;
s_fMaterials[4] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[5] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[6] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[7] = ((data)&0xFF)/255.0f;
}
break;
case 0x100c: //GXSetChanMatcolor (rgba)
if (xfregs.colChans[0].matColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 4;
xfregs.colChans[0].matColor = data;
s_fMaterials[8] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[9] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[10] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[11] = ((data)&0xFF)/255.0f;
}
break;
case 0x100d: //GXSetChanMatcolor (rgba)
if (xfregs.colChans[1].matColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 8;
xfregs.colChans[1].matColor = data;
s_fMaterials[12] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[13] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[14] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[15] = ((data)&0xFF)/255.0f;
}
break;
case 0x100e: // color0
if (xfregs.colChans[0].color.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[0].color.hex = data;
}
break;
case 0x100f: // color1
if (xfregs.colChans[1].color.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[1].color.hex = data;
}
break;
case 0x1010: // alpha0
if (xfregs.colChans[0].alpha.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[0].alpha.hex = data;
}
break;
case 0x1011: // alpha1
if (xfregs.colChans[1].alpha.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[1].alpha.hex = data;
}
break;
case 0x1012: // dual tex transform
if (xfregs.bEnableDualTexTransform != (data&1)) {
VertexManager::Flush();
xfregs.bEnableDualTexTransform = data&1;
}
break;
case 0x1013:
case 0x1014:
case 0x1015:
case 0x1016:
case 0x1017:
DEBUG_LOG("xf addr: %x=%x\n", address, data);
break;
case 0x1018:
//_assert_msg_(GX_XF, 0, "XF matrixindex0");
VertexShaderMngr::SetTexMatrixChangedA(data); //?
break;
case 0x1019:
//_assert_msg_(GX_XF, 0, "XF matrixindex1");
VertexShaderMngr::SetTexMatrixChangedB(data); //?
break;
case 0x101a:
VertexManager::Flush();
VertexShaderMngr::SetViewport((float*)&pData[i]);
i += 6;
break;
case 0x101c: // paper mario writes 16777216.0f, 1677721.75
break;
case 0x101f: // paper mario writes 16777216.0f, 5033165.0f
break;
case 0x1020:
VertexManager::Flush();
VertexShaderMngr::SetProjection((float*)&pData[i]);
i += 7;
return;
case 0x103f: // GXSetNumTexGens
if (xfregs.numTexGens != data) {
VertexManager::Flush();
xfregs.numTexGens = data;
}
break;
case 0x1040: xfregs.texcoords[0].texmtxinfo.hex = data; break;
case 0x1041: xfregs.texcoords[1].texmtxinfo.hex = data; break;
case 0x1042: xfregs.texcoords[2].texmtxinfo.hex = data; break;
case 0x1043: xfregs.texcoords[3].texmtxinfo.hex = data; break;
case 0x1044: xfregs.texcoords[4].texmtxinfo.hex = data; break;
case 0x1045: xfregs.texcoords[5].texmtxinfo.hex = data; break;
case 0x1046: xfregs.texcoords[6].texmtxinfo.hex = data; break;
case 0x1047: xfregs.texcoords[7].texmtxinfo.hex = data; break;
case 0x1048:
case 0x1049:
case 0x104a:
case 0x104b:
case 0x104c:
case 0x104d:
case 0x104e:
case 0x104f:
DEBUG_LOG("xf addr: %x=%x\n", address, data);
break;
case 0x1050: xfregs.texcoords[0].postmtxinfo.hex = data; break;
case 0x1051: xfregs.texcoords[1].postmtxinfo.hex = data; break;
case 0x1052: xfregs.texcoords[2].postmtxinfo.hex = data; break;
case 0x1053: xfregs.texcoords[3].postmtxinfo.hex = data; break;
case 0x1054: xfregs.texcoords[4].postmtxinfo.hex = data; break;
case 0x1055: xfregs.texcoords[5].postmtxinfo.hex = data; break;
case 0x1056: xfregs.texcoords[6].postmtxinfo.hex = data; break;
case 0x1057: xfregs.texcoords[7].postmtxinfo.hex = data; break;
default:
DEBUG_LOG("xf addr: %x=%x\n", address, data);
break;
}
}
else if (address>=0x4000)
{
// MessageBox(NULL, "1", "1", MB_OK);
//4010 __GXSetGenMode
}
}
}
// Check docs for this sucker...
void VertexShaderMngr::LoadIndexedXF(u32 val, int array)
{
int index = val>>16;
int address = val&0xFFF; //check mask
int size = ((val>>12)&0xF)+1;
//load stuff from array to address in xf mem
VertexManager::Flush();
InvalidateXFRange(address, address+size);
//PRIM_LOG("xfmem iwrite: 0x%x-0x%x\n", address, address+size);
for (int i = 0; i < size; i++)
xfmem[address + i] = Memory_Read_U32(arraybases[array] + arraystrides[array]*index + i*4);
}
float* VertexShaderMngr::GetPosNormalMat()
{
return (float*)xfmem + MatrixIndexA.PosNormalMtxIdx * 4;
}
void VertexShaderMngr::GetVertexShaderId(VERTEXSHADERUID& id, u32 components)
{
u32 zbufrender = (bpmem.ztex2.op==ZTEXTURE_ADD)||Renderer::GetZBufferTarget()!=0;
id.values[0] = components|(xfregs.numTexGens<<23)|(xfregs.nNumChans<<27)|((u32)xfregs.bEnableDualTexTransform<<29)|(zbufrender<<30);
for(int i = 0; i < 2; ++i) {
id.values[1+i] = xfregs.colChans[i].color.enablelighting?(u32)xfregs.colChans[i].color.hex:(u32)xfregs.colChans[i].color.matsource;
id.values[1+i] |= (xfregs.colChans[i].alpha.enablelighting?(u32)xfregs.colChans[i].alpha.hex:(u32)xfregs.colChans[i].alpha.matsource)<<15;
}
// fog
id.values[1] |= (((u32)bpmem.fog.c_proj_fsel.fsel&3)<<30);
id.values[2] |= (((u32)bpmem.fog.c_proj_fsel.fsel>>2)<<30);
u32* pcurvalue = &id.values[3];
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;
u32 val = ((tinfo.hex>>1)&0x1ffff);
if( xfregs.bEnableDualTexTransform && tinfo.texgentype == XF_TEXGEN_REGULAR ) {
// rewrite normalization and post index
val |= ((u32)xfregs.texcoords[i].postmtxinfo.index<<17)|((u32)xfregs.texcoords[i].postmtxinfo.normalize<<23);
}
switch(i & 3) {
case 0: pcurvalue[0] |= val; break;
case 1: pcurvalue[0] |= val<<24; pcurvalue[1] = val>>8; ++pcurvalue; break;
case 2: pcurvalue[0] |= val<<16; pcurvalue[1] = val>>16; ++pcurvalue; break;
case 3: pcurvalue[0] |= val<<8; ++pcurvalue; break;
}
}
}
static char text[16384];
#define WRITE p+=sprintf
#define LIGHTS_POS ""
bool VertexShaderMngr::GenerateVertexShader(VERTEXSHADER& vs, u32 components)
char* GenerateLightShader(char* p, int index, const LitChannel& chan, const char* dest, int coloralpha);
bool GenerateVertexShader(VERTEXSHADER& vs, u32 components)
{
DVSTARTPROFILE();
@ -1180,7 +408,7 @@ bool VertexShaderMngr::GenerateVertexShader(VERTEXSHADER& vs, u32 components)
}
// coloralpha - 1 if color, 2 if alpha
char* VertexShaderMngr::GenerateLightShader(char* p, int index, const LitChannel& chan, const char* dest, int coloralpha)
char* GenerateLightShader(char* p, int index, const LitChannel& chan, const char* dest, int coloralpha)
{
const char* swizzle = "xyzw";
if( coloralpha == 1 ) swizzle = "xyz";

95
Source/Plugins/Plugin_VideoOGL/Src/VertexShader.h
View File

@ -18,7 +18,6 @@
#ifndef GCOGL_VERTEXSHADER_H
#define GCOGL_VERTEXSHADER_H
#include <map>
/////////////
// Lighting
@ -226,99 +225,7 @@ struct VERTEXSHADER
#endif
};
class VertexShaderMngr
{
struct VSCacheEntry
{
VERTEXSHADER shader;
int frameCount;
VSCacheEntry() : frameCount(0) {}
void Destroy() {
SAFE_RELEASE_PROG(shader.glprogid);
}
};
class VERTEXSHADERUID
{
public:
VERTEXSHADERUID() {}
VERTEXSHADERUID(const VERTEXSHADERUID& r) {
for(size_t i = 0; i < sizeof(values) / sizeof(u32); ++i)
values[i] = r.values[i];
}
bool operator<(const VERTEXSHADERUID& _Right) const
{
if( values[0] < _Right.values[0] )
return true;
else if( values[0] > _Right.values[0] )
return false;
int N = (((values[0]>>23)&0xf)*3+3)/4 + 3; // numTexGens*3/4+1
for(int i = 1; i < N; ++i) {
if( values[i] < _Right.values[i] )
return true;
else if( values[i] > _Right.values[i] )
return false;
}
return false;
}
bool operator==(const VERTEXSHADERUID& _Right) const
{
if( values[0] != _Right.values[0] )
return false;
int N = (((values[0]>>23)&0xf)*3+3)/4 + 3; // numTexGens*3/4+1
for(int i = 1; i < N; ++i) {
if( values[i] != _Right.values[i] )
return false;
}
return true;
}
u32 values[9];
};
typedef std::map<VERTEXSHADERUID,VSCacheEntry> VSCache;
static VSCache vshaders;
static VERTEXSHADER* pShaderLast;
static TMatrixIndexA MatrixIndexA;
static TMatrixIndexB MatrixIndexB;
static void GetVertexShaderId(VERTEXSHADERUID& uid, u32 components);
static bool GenerateVertexShader(VERTEXSHADER& vs, u32 components);
static char* GenerateLightShader(char* p, int index, const LitChannel& chan, const char* dest, int coloralpha);
public:
static void Init();
static void Cleanup();
static void Shutdown();
static VERTEXSHADER* GetShader(u32 components);
static bool CompileVertexShader(VERTEXSHADER& ps, const char* pstrprogram);
// constant management
static void SetConstants(VERTEXSHADER& vs);
static void SetViewport(float* _Viewport);
static void SetViewportChanged();
static void SetProjection(float* _pProjection, int constantIndex = -1);
static void InvalidateXFRange(int start, int end);
static void SetTexMatrixChangedA(u32 Value);
static void SetTexMatrixChangedB(u32 Value);
static size_t SaveLoadState(char *ptr, BOOL save);
static void LoadXFReg(u32 transferSize, u32 address, u32 *pData);
static void LoadIndexedXF(u32 val, int array);
static float* GetPosNormalMat();
static float rawViewport[6];
static float rawProjection[7];
};
bool GenerateVertexShader(VERTEXSHADER& vs, u32 components);
extern XFRegisters xfregs;

781
Source/Plugins/Plugin_VideoOGL/Src/VertexShaderManager.cpp Normal file
View File

@ -0,0 +1,781 @@
// 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 <math.h>
#include "Render.h"
#include "VertexShader.h"
#include "VertexShaderManager.h"
#include "BPStructs.h"
#include "VertexLoader.h"
VertexShaderMngr::VSCache VertexShaderMngr::vshaders;
VERTEXSHADER* VertexShaderMngr::pShaderLast = NULL;
TMatrixIndexA VertexShaderMngr::MatrixIndexA;
TMatrixIndexB VertexShaderMngr::MatrixIndexB;
float VertexShaderMngr::rawViewport[6] = {0};
float VertexShaderMngr::rawProjection[7] = {0};
float GC_ALIGNED16(g_fProjectionMatrix[16]);
static int s_nMaxVertexInstructions;
////////////////////////
// Internal Variables //
////////////////////////
XFRegisters xfregs;
static u32 xfmem[XFMEM_SIZE];
static float s_fMaterials[16];
// track changes
static bool bTexMatricesChanged[2], bPosNormalMatrixChanged, bProjectionChanged, bViewportChanged;
int nMaterialsChanged;
static int nTransformMatricesChanged[2]; // min,max
static int nNormalMatricesChanged[2]; // min,max
static int nPostTransformMatricesChanged[2]; // min,max
static int nLightsChanged[2]; // min,max
void VertexShaderMngr::Init()
{
nTransformMatricesChanged[0] = nTransformMatricesChanged[1] = -1;
nNormalMatricesChanged[0] = nNormalMatricesChanged[1] = -1;
nPostTransformMatricesChanged[0] = nPostTransformMatricesChanged[1] = -1;
nLightsChanged[0] = nLightsChanged[1] = -1;
bTexMatricesChanged[0] = bTexMatricesChanged[1] = false;
bPosNormalMatrixChanged = bProjectionChanged = bViewportChanged = false;
nMaterialsChanged = 0;
memset(&xfregs, 0, sizeof(xfregs));
memset(xfmem, 0, sizeof(xfmem));
glGetProgramivARB(GL_VERTEX_PROGRAM_ARB, GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB, &s_nMaxVertexInstructions);
}
void VertexShaderMngr::Shutdown()
{
VSCache::iterator iter = vshaders.begin();
for (;iter!=vshaders.end();iter++)
iter->second.Destroy();
vshaders.clear();
}
VERTEXSHADER* VertexShaderMngr::GetShader(u32 components)
{
DVSTARTPROFILE();
VERTEXSHADERUID uid;
GetVertexShaderId(uid, components);
VSCache::iterator iter = vshaders.find(uid);
if (iter != vshaders.end()) {
iter->second.frameCount=frameCount;
VSCacheEntry &entry = iter->second;
if (&entry.shader != pShaderLast) {
pShaderLast = &entry.shader;
}
return pShaderLast;
}
VSCacheEntry& entry = vshaders[uid];
if (!GenerateVertexShader(entry.shader, components)) {
ERROR_LOG("failed to create vertex shader\n");
}
//Make an entry in the table
entry.frameCount=frameCount;
pShaderLast = &entry.shader;
INCSTAT(stats.numVertexShadersCreated);
SETSTAT(stats.numVertexShadersAlive,vshaders.size());
return pShaderLast;
}
void VertexShaderMngr::Cleanup()
{
VSCache::iterator iter=vshaders.begin();
while (iter != vshaders.end()) {
VSCacheEntry &entry = iter->second;
if (entry.frameCount < frameCount-200) {
entry.Destroy();
#ifdef _WIN32
iter = vshaders.erase(iter);
#else
vshaders.erase(iter++);
#endif
}
else {
++iter;
}
}
// static int frame = 0;
// if( frame++ > 30 ) {
// VSCache::iterator iter=vshaders.begin();
// while(iter!=vshaders.end()) {
// iter->second.Destroy();
// ++iter;
// }
// vshaders.clear();
// }
SETSTAT(stats.numPixelShadersAlive,vshaders.size());
}
bool VertexShaderMngr::CompileVertexShader(VERTEXSHADER& vs, const char* pstrprogram)
{
char stropt[64];
sprintf(stropt, "MaxLocalParams=256,MaxInstructions=%d", s_nMaxVertexInstructions);
#ifdef _WIN32
const char* opts[] = {"-profileopts",stropt,"-O2", "-q", NULL};
#else
const char* opts[] = {"-profileopts",stropt,"-q", NULL};
#endif
CGprogram tempprog = cgCreateProgram(g_cgcontext, CG_SOURCE, pstrprogram, g_cgvProf, "main", opts);
if (!cgIsProgram(tempprog) || cgGetError() != CG_NO_ERROR) {
ERROR_LOG("Failed to load vs %s:\n", cgGetLastListing(g_cgcontext));
ERROR_LOG(pstrprogram);
return false;
}
//ERROR_LOG(pstrprogram);
//ERROR_LOG("id: %d\n", g_Config.iSaveTargetId);
char* pcompiledprog = (char*)cgGetProgramString(tempprog, CG_COMPILED_PROGRAM);
char* plocal = strstr(pcompiledprog, "program.local");
while( plocal != NULL ) {
const char* penv = " program.env";
memcpy(plocal, penv, 13);
plocal = strstr(plocal+13, "program.local");
}
glGenProgramsARB( 1, &vs.glprogid );
glBindProgramARB( GL_VERTEX_PROGRAM_ARB, vs.glprogid );
glProgramStringARB( GL_VERTEX_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, (GLsizei)strlen(pcompiledprog), pcompiledprog);
GLenum err = GL_NO_ERROR;
GL_REPORT_ERROR();
if( err != GL_NO_ERROR ) {
ERROR_LOG(pstrprogram);
ERROR_LOG(pcompiledprog);
}
cgDestroyProgram(tempprog);
#ifdef _DEBUG
vs.strprog = pstrprogram;
#endif
return true;
}
// TODO: this array is misdeclared. Why teh f** does it go through the compilers?
const u16 s_mtrltable[16][2] = {0, 0, 0, 1, 1, 1, 0, 2,
2, 1, 0, 3, 1, 2, 0, 3,
3, 1, 0, 4, 1, 3, 0, 4,
2, 2, 0, 4, 1, 3, 0, 4};
/// syncs the shader constant buffers with xfmem
void VertexShaderMngr::SetConstants(VERTEXSHADER& vs)
{
//nTransformMatricesChanged[0] = 0; nTransformMatricesChanged[1] = 256;
//nNormalMatricesChanged[0] = 0; nNormalMatricesChanged[1] = 96;
//nPostTransformMatricesChanged[0] = 0; nPostTransformMatricesChanged[1] = 256;
//nLightsChanged[0] = 0; nLightsChanged[1] = 0x80;
//bPosNormalMatrixChanged = true;
//bTexMatricesChanged[0] = bTexMatricesChanged[1] = true;
//bProjectionChanged = true;
// bPosNormalMatrixChanged = bTexMatricesChanged[0] = bTexMatricesChanged[1] = true;
// nMaterialsChanged = 15;
if (nTransformMatricesChanged[0] >= 0) {
int startn = nTransformMatricesChanged[0]/4;
int endn = (nTransformMatricesChanged[1]+3)/4;
const float* pstart = (const float*)&xfmem[startn*4];
for(int i = startn; i < endn; ++i, pstart += 4)
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TRANSFORMMATRICES+i, pstart);
nTransformMatricesChanged[0] = nTransformMatricesChanged[1] = -1;
}
if (nNormalMatricesChanged[0] >= 0) {
int startn = nNormalMatricesChanged[0]/3;
int endn = (nNormalMatricesChanged[1]+2)/3;
const float* pnstart = (const float*)&xfmem[XFMEM_NORMALMATRICES+3*startn];
for(int i = startn; i < endn; ++i, pnstart += 3)
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_NORMALMATRICES+i, pnstart);
nNormalMatricesChanged[0] = nNormalMatricesChanged[1] = -1;
}
if (nPostTransformMatricesChanged[0] >= 0) {
int startn = nPostTransformMatricesChanged[0]/4;
int endn = (nPostTransformMatricesChanged[1]+3)/4;
const float* pstart = (const float*)&xfmem[XFMEM_POSTMATRICES+startn*4];
for(int i = startn; i < endn; ++i, pstart += 4)
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSTTRANSFORMMATRICES+i, pstart);
}
if (nLightsChanged[0] >= 0) {
// lights don't have a 1 to 1 mapping, the color component needs to be converted to 4 floats
int istart = nLightsChanged[0]/0x10;
int iend = (nLightsChanged[1]+15)/0x10;
const float* xfmemptr = (const float*)&xfmem[0x10*istart+XFMEM_LIGHTS];
for(int i = istart; i < iend; ++i) {
u32 color = *(const u32*)(xfmemptr+3);
glProgramEnvParameter4fARB(GL_VERTEX_PROGRAM_ARB, C_LIGHTS+5*i,
((color>>24)&0xFF)/255.0f, ((color>>16)&0xFF)/255.0f, ((color>>8)&0xFF)/255.0f, ((color)&0xFF)/255.0f);
xfmemptr += 4;
for(int j = 0; j < 4; ++j, xfmemptr += 3) {
if( j == 1 && fabs(xfmemptr[0]) < 0.00001f && fabs(xfmemptr[1]) < 0.00001f && fabs(xfmemptr[2]) < 0.00001f) {
// dist atten, make sure not equal to 0!!!
glProgramEnvParameter4fARB(GL_VERTEX_PROGRAM_ARB, C_LIGHTS+5*i+j+1, 0.00001f, xfmemptr[1], xfmemptr[2], 0);
}
else
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_LIGHTS+5*i+j+1, xfmemptr);
}
}
nLightsChanged[0] = nLightsChanged[1] = -1;
}
if (nMaterialsChanged) {
for(int i = 0; i < 4; ++i) {
if( nMaterialsChanged&(1<<i) )
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_MATERIALS+i, &s_fMaterials[4*i]);
}
nMaterialsChanged = 0;
}
if (bPosNormalMatrixChanged) {
bPosNormalMatrixChanged = false;
float* pos = (float*)xfmem + MatrixIndexA.PosNormalMtxIdx * 4;
float* norm = (float*)xfmem + XFMEM_NORMALMATRICES + 3 * (MatrixIndexA.PosNormalMtxIdx & 31);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX, pos);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+1, pos+4);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+2, pos+8);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+3, norm);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+4, norm+3);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_POSNORMALMATRIX+5, norm+6);
}
if (bTexMatricesChanged[0]) {
bTexMatricesChanged[0] = false;
float* fptrs[] = {(float*)xfmem + MatrixIndexA.Tex0MtxIdx * 4, (float*)xfmem + MatrixIndexA.Tex1MtxIdx * 4,
(float*)xfmem + MatrixIndexA.Tex2MtxIdx * 4, (float*)xfmem + MatrixIndexA.Tex3MtxIdx * 4 };
for(int i = 0; i < 4; ++i) {
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i, fptrs[i]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+1, fptrs[i]+4);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+2, fptrs[i]+8);
}
}
if (bTexMatricesChanged[1]) {
bTexMatricesChanged[1] = false;
float* fptrs[] = {(float*)xfmem + MatrixIndexB.Tex4MtxIdx * 4, (float*)xfmem + MatrixIndexB.Tex5MtxIdx * 4,
(float*)xfmem + MatrixIndexB.Tex6MtxIdx * 4, (float*)xfmem + MatrixIndexB.Tex7MtxIdx * 4 };
for(int i = 0; i < 4; ++i) {
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+12, fptrs[i]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+12+1, fptrs[i]+4);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_TEXMATRICES+3*i+12+2, fptrs[i]+8);
}
}
if (bViewportChanged) {
bViewportChanged = false;
// reversed gxsetviewport(xorig, yorig, width, height, nearz, farz)
// [0] = width/2
// [1] = height/2
// [2] = 16777215 * (farz-nearz)
// [3] = xorig + width/2 + 342
// [4] = yorig + height/2 + 342
// [5] = 16777215 * farz
INFO_LOG("view: topleft=(%f,%f), wh=(%f,%f), z=(%f,%f)\n",rawViewport[3]-rawViewport[0]-342,rawViewport[4]+rawViewport[1]-342,
2 * rawViewport[0], 2 * rawViewport[1], (rawViewport[5]-rawViewport[2])/16777215.0f, rawViewport[5]/16777215.0f);
glViewport((int)(rawViewport[3]-rawViewport[0]-342)<<g_AAx,Renderer::GetTargetHeight()-((int)(rawViewport[4]-rawViewport[1]-342)<<g_AAy), abs((int)(2 * rawViewport[0])), abs((int)(2 * rawViewport[1])));
glDepthRange((rawViewport[5]-rawViewport[2])/16777215.0f, rawViewport[5]/16777215.0f);
}
if (bProjectionChanged) {
bProjectionChanged = false;
if (rawProjection[6] == 0) {
g_fProjectionMatrix[0] = rawProjection[0];
g_fProjectionMatrix[1] = 0.0f;
g_fProjectionMatrix[2] = rawProjection[1];
g_fProjectionMatrix[3] = 0;//-0.5f/Renderer::GetTargetWidth();
g_fProjectionMatrix[4] = 0.0f;
g_fProjectionMatrix[5] = rawProjection[2];
g_fProjectionMatrix[6] = rawProjection[3];
g_fProjectionMatrix[7] = 0;//+0.5f/Renderer::GetTargetHeight();
g_fProjectionMatrix[8] = 0.0f;
g_fProjectionMatrix[9] = 0.0f;
g_fProjectionMatrix[10] = -(1-rawProjection[4]);
g_fProjectionMatrix[11] = rawProjection[5];
g_fProjectionMatrix[12] = 0.0f;
g_fProjectionMatrix[13] = 0.0f;
g_fProjectionMatrix[14] = -1.0f;
g_fProjectionMatrix[15] = 0.0f;
}
else {
g_fProjectionMatrix[0] = rawProjection[0];
g_fProjectionMatrix[1] = 0.0f;
g_fProjectionMatrix[2] = 0.0f;
g_fProjectionMatrix[3] = rawProjection[1];
g_fProjectionMatrix[4] = 0.0f;
g_fProjectionMatrix[5] = rawProjection[2];
g_fProjectionMatrix[6] = 0.0f;
g_fProjectionMatrix[7] = rawProjection[3];
g_fProjectionMatrix[8] = 0.0f;
g_fProjectionMatrix[9] = 0.0f;
g_fProjectionMatrix[10] = rawProjection[4];
g_fProjectionMatrix[11] = -(-1 - rawProjection[5]);
g_fProjectionMatrix[12] = 0;
g_fProjectionMatrix[13] = 0;
g_fProjectionMatrix[14] = 0.0f;
g_fProjectionMatrix[15] = 1.0f;
}
PRIM_LOG("Projection: %f %f %f %f %f %f\n",rawProjection[0], rawProjection[1], rawProjection[2], rawProjection[3], rawProjection[4], rawProjection[5]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION, &g_fProjectionMatrix[0]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION+1, &g_fProjectionMatrix[4]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION+2, &g_fProjectionMatrix[8]);
glProgramEnvParameter4fvARB(GL_VERTEX_PROGRAM_ARB, C_PROJECTION+3, &g_fProjectionMatrix[12]);
}
}
void VertexShaderMngr::InvalidateXFRange(int start, int end)
{
if( ((u32)start >= MatrixIndexA.PosNormalMtxIdx*4 && (u32)start < MatrixIndexA.PosNormalMtxIdx*4+12) ||
((u32)start >= XFMEM_NORMALMATRICES+(MatrixIndexA.PosNormalMtxIdx&31)*3 && (u32)start < XFMEM_NORMALMATRICES+(MatrixIndexA.PosNormalMtxIdx&31)*3+9) ) {
bPosNormalMatrixChanged = true;
}
if (((u32)start >= MatrixIndexA.Tex0MtxIdx*4 && (u32)start < MatrixIndexA.Tex0MtxIdx*4+12) ||
((u32)start >= MatrixIndexA.Tex1MtxIdx*4 && (u32)start < MatrixIndexA.Tex1MtxIdx*4+12) ||
((u32)start >= MatrixIndexA.Tex2MtxIdx*4 && (u32)start < MatrixIndexA.Tex2MtxIdx*4+12) ||
((u32)start >= MatrixIndexA.Tex3MtxIdx*4 && (u32)start < MatrixIndexA.Tex3MtxIdx*4+12)) {
bTexMatricesChanged[0] = true;
}
if (((u32)start >= MatrixIndexB.Tex4MtxIdx*4 && (u32)start < MatrixIndexB.Tex4MtxIdx*4+12) ||
((u32)start >= MatrixIndexB.Tex5MtxIdx*4 && (u32)start < MatrixIndexB.Tex5MtxIdx*4+12) ||
((u32)start >= MatrixIndexB.Tex6MtxIdx*4 && (u32)start < MatrixIndexB.Tex6MtxIdx*4+12) ||
((u32)start >= MatrixIndexB.Tex7MtxIdx*4 && (u32)start < MatrixIndexB.Tex7MtxIdx*4+12)) {
bTexMatricesChanged[1] = true;
}
if (start < XFMEM_POSMATRICES_END ) {
if (nTransformMatricesChanged[0] == -1) {
nTransformMatricesChanged[0] = start;
nTransformMatricesChanged[1] = end>XFMEM_POSMATRICES_END?XFMEM_POSMATRICES_END:end;
}
else {
if (nTransformMatricesChanged[0] > start) nTransformMatricesChanged[0] = start;
if (nTransformMatricesChanged[1] < end) nTransformMatricesChanged[1] = end>XFMEM_POSMATRICES_END?XFMEM_POSMATRICES_END:end;
}
}
if (start < XFMEM_NORMALMATRICES_END && end > XFMEM_NORMALMATRICES ) {
int _start = start < XFMEM_NORMALMATRICES ? 0 : start-XFMEM_NORMALMATRICES;
int _end = end < XFMEM_NORMALMATRICES_END ? end-XFMEM_NORMALMATRICES : XFMEM_NORMALMATRICES_END-XFMEM_NORMALMATRICES;
if (nNormalMatricesChanged[0] == -1 ) {
nNormalMatricesChanged[0] = _start;
nNormalMatricesChanged[1] = _end;
}
else {
if (nNormalMatricesChanged[0] > _start) nNormalMatricesChanged[0] = _start;
if (nNormalMatricesChanged[1] < _end) nNormalMatricesChanged[1] = _end;
}
}
if (start < XFMEM_POSTMATRICES_END && end > XFMEM_POSTMATRICES ) {
int _start = start < XFMEM_POSTMATRICES ? XFMEM_POSTMATRICES : start-XFMEM_POSTMATRICES;
int _end = end < XFMEM_POSTMATRICES_END ? end-XFMEM_POSTMATRICES : XFMEM_POSTMATRICES_END-XFMEM_POSTMATRICES;
if (nPostTransformMatricesChanged[0] == -1 ) {
nPostTransformMatricesChanged[0] = _start;
nPostTransformMatricesChanged[1] = _end;
}
else {
if (nPostTransformMatricesChanged[0] > _start) nPostTransformMatricesChanged[0] = _start;
if (nPostTransformMatricesChanged[1] < _end) nPostTransformMatricesChanged[1] = _end;
}
}
if (start < XFMEM_LIGHTS_END && end > XFMEM_LIGHTS) {
int _start = start < XFMEM_LIGHTS ? XFMEM_LIGHTS : start-XFMEM_LIGHTS;
int _end = end < XFMEM_LIGHTS_END ? end-XFMEM_LIGHTS : XFMEM_LIGHTS_END-XFMEM_LIGHTS;
if (nLightsChanged[0] == -1 ) {
nLightsChanged[0] = _start;
nLightsChanged[1] = _end;
}
else {
if (nLightsChanged[0] > _start) nLightsChanged[0] = _start;
if (nLightsChanged[1] < _end) nLightsChanged[1] = _end;
}
}
}
void VertexShaderMngr::SetTexMatrixChangedA(u32 Value)
{
if (MatrixIndexA.Hex != Value) {
VertexManager::Flush();
if (MatrixIndexA.PosNormalMtxIdx != (Value&0x3f))
bPosNormalMatrixChanged = true;
bTexMatricesChanged[0] = true;
MatrixIndexA.Hex = Value;
}
}
void VertexShaderMngr::SetTexMatrixChangedB(u32 Value)
{
if (MatrixIndexB.Hex != Value) {
VertexManager::Flush();
bTexMatricesChanged[1] = true;
MatrixIndexB.Hex = Value;
}
}
void VertexShaderMngr::SetViewport(float* _Viewport)
{
// check for paper mario
for (size_t i = 0; i < ARRAYSIZE(rawViewport); ++i) {
if( *(u32*)(_Viewport + i) == 0x7f800000 )
return; // invalid number
}
memcpy(rawViewport, _Viewport, sizeof(rawViewport));
bViewportChanged = true;
}
void VertexShaderMngr::SetViewportChanged()
{
bViewportChanged = true;
}
void VertexShaderMngr::SetProjection(float* _pProjection, int constantIndex)
{
memcpy(rawProjection, _pProjection, sizeof(rawProjection));
bProjectionChanged = true;
}
size_t VertexShaderMngr::SaveLoadState(char *ptr, BOOL save)
{
BEGINSAVELOAD;
SAVELOAD(&xfregs,sizeof(xfregs));
SAVELOAD(xfmem,XFMEM_SIZE*sizeof(u32));
SAVELOAD(rawViewport,sizeof(rawViewport));
SAVELOAD(rawProjection,sizeof(rawProjection));
SAVELOAD(&MatrixIndexA,sizeof(TMatrixIndexA));
SAVELOAD(&MatrixIndexB,sizeof(TMatrixIndexB));
if (!save) {
// invalidate all
InvalidateXFRange(0,0x1000);
}
ENDSAVELOAD;
}
// LoadXFReg 0x10
void VertexShaderMngr::LoadXFReg(u32 transferSize, u32 baseAddress, u32 *pData)
{
u32 address = baseAddress;
for (int i=0; i<(int)transferSize; i++)
{
address = baseAddress + i;
// Setup a Matrix
if (address < 0x1000)
{
VertexManager::Flush();
InvalidateXFRange(address, address+transferSize);
//PRIM_LOG("xfmem write: 0x%x-0x%x\n", address, address+transferSize);
u32* p1 = &xfmem[address];
memcpy_gc(p1, &pData[i], transferSize*4);
i += transferSize;
}
else if (address<0x2000)
{
u32 data = pData[i];
switch (address)
{
case 0x1000: // error
break;
case 0x1001: // diagnostics
break;
case 0x1002: // internal state 0
break;
case 0x1003: // internal state 1
break;
case 0x1004: // xf_clock
break;
case 0x1005: // clipdisable
if (data & 1) { // disable clipping detection
}
if (data & 2) { // disable trivial rejection
}
if (data & 4) { // disable cpoly clipping acceleration
}
break;
case 0x1006: //SetGPMetric
break;
case 0x1008: //__GXXfVtxSpecs, wrote 0004
xfregs.hostinfo = *(INVTXSPEC*)&data;
break;
case 0x1009: //GXSetNumChans (no)
if (xfregs.nNumChans != (data&3) ) {
VertexManager::Flush();
xfregs.nNumChans = data&3;
}
break;
case 0x100a: //GXSetChanAmbientcolor
if (xfregs.colChans[0].ambColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 1;
xfregs.colChans[0].ambColor = data;
s_fMaterials[0] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[1] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[2] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[3] = ((data)&0xFF)/255.0f;
}
break;
case 0x100b: //GXSetChanAmbientcolor
if (xfregs.colChans[1].ambColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 2;
xfregs.colChans[1].ambColor = data;
s_fMaterials[4] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[5] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[6] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[7] = ((data)&0xFF)/255.0f;
}
break;
case 0x100c: //GXSetChanMatcolor (rgba)
if (xfregs.colChans[0].matColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 4;
xfregs.colChans[0].matColor = data;
s_fMaterials[8] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[9] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[10] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[11] = ((data)&0xFF)/255.0f;
}
break;
case 0x100d: //GXSetChanMatcolor (rgba)
if (xfregs.colChans[1].matColor != data) {
VertexManager::Flush();
nMaterialsChanged |= 8;
xfregs.colChans[1].matColor = data;
s_fMaterials[12] = ((data>>24)&0xFF)/255.0f;
s_fMaterials[13] = ((data>>16)&0xFF)/255.0f;
s_fMaterials[14] = ((data>>8)&0xFF)/255.0f;
s_fMaterials[15] = ((data)&0xFF)/255.0f;
}
break;
case 0x100e: // color0
if (xfregs.colChans[0].color.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[0].color.hex = data;
}
break;
case 0x100f: // color1
if (xfregs.colChans[1].color.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[1].color.hex = data;
}
break;
case 0x1010: // alpha0
if (xfregs.colChans[0].alpha.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[0].alpha.hex = data;
}
break;
case 0x1011: // alpha1
if (xfregs.colChans[1].alpha.hex != (data&0x7fff) ) {
VertexManager::Flush();
xfregs.colChans[1].alpha.hex = data;
}
break;
case 0x1012: // dual tex transform
if (xfregs.bEnableDualTexTransform != (data&1)) {
VertexManager::Flush();
xfregs.bEnableDualTexTransform = data&1;
}
break;
case 0x1013:
case 0x1014:
case 0x1015:
case 0x1016:
case 0x1017:
DEBUG_LOG("xf addr: %x=%x\n", address, data);
break;
case 0x1018:
//_assert_msg_(GX_XF, 0, "XF matrixindex0");
VertexShaderMngr::SetTexMatrixChangedA(data); //?
break;
case 0x1019:
//_assert_msg_(GX_XF, 0, "XF matrixindex1");
VertexShaderMngr::SetTexMatrixChangedB(data); //?
break;
case 0x101a:
VertexManager::Flush();
VertexShaderMngr::SetViewport((float*)&pData[i]);
i += 6;
break;
case 0x101c: // paper mario writes 16777216.0f, 1677721.75
break;
case 0x101f: // paper mario writes 16777216.0f, 5033165.0f
break;
case 0x1020:
VertexManager::Flush();
VertexShaderMngr::SetProjection((float*)&pData[i]);
i += 7;
return;
case 0x103f: // GXSetNumTexGens
if (xfregs.numTexGens != data) {
VertexManager::Flush();
xfregs.numTexGens = data;
}
break;
case 0x1040: xfregs.texcoords[0].texmtxinfo.hex = data; break;
case 0x1041: xfregs.texcoords[1].texmtxinfo.hex = data; break;
case 0x1042: xfregs.texcoords[2].texmtxinfo.hex = data; break;
case 0x1043: xfregs.texcoords[3].texmtxinfo.hex = data; break;
case 0x1044: xfregs.texcoords[4].texmtxinfo.hex = data; break;
case 0x1045: xfregs.texcoords[5].texmtxinfo.hex = data; break;
case 0x1046: xfregs.texcoords[6].texmtxinfo.hex = data; break;
case 0x1047: xfregs.texcoords[7].texmtxinfo.hex = data; break;
case 0x1048:
case 0x1049:
case 0x104a:
case 0x104b:
case 0x104c:
case 0x104d:
case 0x104e:
case 0x104f:
DEBUG_LOG("xf addr: %x=%x\n", address, data);
break;
case 0x1050: xfregs.texcoords[0].postmtxinfo.hex = data; break;
case 0x1051: xfregs.texcoords[1].postmtxinfo.hex = data; break;
case 0x1052: xfregs.texcoords[2].postmtxinfo.hex = data; break;
case 0x1053: xfregs.texcoords[3].postmtxinfo.hex = data; break;
case 0x1054: xfregs.texcoords[4].postmtxinfo.hex = data; break;
case 0x1055: xfregs.texcoords[5].postmtxinfo.hex = data; break;
case 0x1056: xfregs.texcoords[6].postmtxinfo.hex = data; break;
case 0x1057: xfregs.texcoords[7].postmtxinfo.hex = data; break;
default:
DEBUG_LOG("xf addr: %x=%x\n", address, data);
break;
}
}
else if (address>=0x4000)
{
// MessageBox(NULL, "1", "1", MB_OK);
//4010 __GXSetGenMode
}
}
}
// Check docs for this sucker...
void VertexShaderMngr::LoadIndexedXF(u32 val, int array)
{
int index = val>>16;
int address = val&0xFFF; //check mask
int size = ((val>>12)&0xF)+1;
//load stuff from array to address in xf mem
VertexManager::Flush();
InvalidateXFRange(address, address+size);
//PRIM_LOG("xfmem iwrite: 0x%x-0x%x\n", address, address+size);
for (int i = 0; i < size; i++)
xfmem[address + i] = Memory_Read_U32(arraybases[array] + arraystrides[array]*index + i*4);
}
float* VertexShaderMngr::GetPosNormalMat()
{
return (float*)xfmem + MatrixIndexA.PosNormalMtxIdx * 4;
}
void VertexShaderMngr::GetVertexShaderId(VERTEXSHADERUID& id, u32 components)
{
u32 zbufrender = (bpmem.ztex2.op==ZTEXTURE_ADD)||Renderer::GetZBufferTarget()!=0;
id.values[0] = components|(xfregs.numTexGens<<23)|(xfregs.nNumChans<<27)|((u32)xfregs.bEnableDualTexTransform<<29)|(zbufrender<<30);
for(int i = 0; i < 2; ++i) {
id.values[1+i] = xfregs.colChans[i].color.enablelighting?(u32)xfregs.colChans[i].color.hex:(u32)xfregs.colChans[i].color.matsource;
id.values[1+i] |= (xfregs.colChans[i].alpha.enablelighting?(u32)xfregs.colChans[i].alpha.hex:(u32)xfregs.colChans[i].alpha.matsource)<<15;
}
// fog
id.values[1] |= (((u32)bpmem.fog.c_proj_fsel.fsel&3)<<30);
id.values[2] |= (((u32)bpmem.fog.c_proj_fsel.fsel>>2)<<30);
u32* pcurvalue = &id.values[3];
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;
u32 val = ((tinfo.hex>>1)&0x1ffff);
if( xfregs.bEnableDualTexTransform && tinfo.texgentype == XF_TEXGEN_REGULAR ) {
// rewrite normalization and post index
val |= ((u32)xfregs.texcoords[i].postmtxinfo.index<<17)|((u32)xfregs.texcoords[i].postmtxinfo.normalize<<23);
}
switch(i & 3) {
case 0: pcurvalue[0] |= val; break;
case 1: pcurvalue[0] |= val<<24; pcurvalue[1] = val>>8; ++pcurvalue; break;
case 2: pcurvalue[0] |= val<<16; pcurvalue[1] = val>>16; ++pcurvalue; break;
case 3: pcurvalue[0] |= val<<8; ++pcurvalue; break;
}
}
}

140
Source/Plugins/Plugin_VideoOGL/Src/VertexShaderManager.h Normal file
View File

@ -0,0 +1,140 @@
// 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 _VERTEXSHADERMANAGER_H
#define _VERTEXSHADERMANAGER_H
#include <map>
#include "VertexShader.h"
// shader variables
#define I_POSNORMALMATRIX "cpnmtx"
#define I_PROJECTION "cproj"
#define I_MATERIALS "cmtrl"
#define I_LIGHTS "clights"
#define I_TEXMATRICES "ctexmtx"
#define I_TRANSFORMMATRICES "ctrmtx"
#define I_NORMALMATRICES "cnmtx"
#define I_POSTTRANSFORMMATRICES "cpostmtx"
#define I_FOGPARAMS "cfog"
#define C_POSNORMALMATRIX 0
#define C_PROJECTION (C_POSNORMALMATRIX+6)
#define C_MATERIALS (C_PROJECTION+4)
#define C_LIGHTS (C_MATERIALS+4)
#define C_TEXMATRICES (C_LIGHTS+40)
#define C_TRANSFORMMATRICES (C_TEXMATRICES+24)
#define C_NORMALMATRICES (C_TRANSFORMMATRICES+64)
#define C_POSTTRANSFORMMATRICES (C_NORMALMATRICES+32)
#define C_FOGPARAMS (C_POSTTRANSFORMMATRICES+64)
class VertexShaderMngr
{
struct VSCacheEntry
{
VERTEXSHADER shader;
int frameCount;
VSCacheEntry() : frameCount(0) {}
void Destroy() {
SAFE_RELEASE_PROG(shader.glprogid);
}
};
class VERTEXSHADERUID
{
public:
VERTEXSHADERUID() {}
VERTEXSHADERUID(const VERTEXSHADERUID& r) {
for(size_t i = 0; i < sizeof(values) / sizeof(u32); ++i)
values[i] = r.values[i];
}
bool operator<(const VERTEXSHADERUID& _Right) const
{
if( values[0] < _Right.values[0] )
return true;
else if( values[0] > _Right.values[0] )
return false;
int N = (((values[0]>>23)&0xf)*3+3)/4 + 3; // numTexGens*3/4+1
for(int i = 1; i < N; ++i) {
if( values[i] < _Right.values[i] )
return true;
else if( values[i] > _Right.values[i] )
return false;
}
return false;
}
bool operator==(const VERTEXSHADERUID& _Right) const
{
if( values[0] != _Right.values[0] )
return false;
int N = (((values[0]>>23)&0xf)*3+3)/4 + 3; // numTexGens*3/4+1
for(int i = 1; i < N; ++i) {
if( values[i] != _Right.values[i] )
return false;
}
return true;
}
u32 values[9];
};
typedef std::map<VERTEXSHADERUID,VSCacheEntry> VSCache;
static VSCache vshaders;
static VERTEXSHADER* pShaderLast;
static TMatrixIndexA MatrixIndexA;
static TMatrixIndexB MatrixIndexB;
static void GetVertexShaderId(VERTEXSHADERUID& uid, u32 components);
public:
static void Init();
static void Cleanup();
static void Shutdown();
static VERTEXSHADER* GetShader(u32 components);
static bool CompileVertexShader(VERTEXSHADER& ps, const char* pstrprogram);
// constant management
static void SetConstants(VERTEXSHADER& vs);
static void SetViewport(float* _Viewport);
static void SetViewportChanged();
static void SetProjection(float* _pProjection, int constantIndex = -1);
static void InvalidateXFRange(int start, int end);
static void SetTexMatrixChangedA(u32 Value);
static void SetTexMatrixChangedB(u32 Value);
static size_t SaveLoadState(char *ptr, BOOL save);
static void LoadXFReg(u32 transferSize, u32 address, u32 *pData);
static void LoadIndexedXF(u32 val, int array);
static float* GetPosNormalMat();
static float rawViewport[6];
static float rawProjection[7];
};
#endif

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

@ -30,8 +30,8 @@
#include "EmuWindow.h"
#endif
#include "VertexLoader.h"
#include "VertexShader.h"
#include "PixelShader.h"
#include "PixelShaderManager.h"
#include "VertexShaderManager.h"
HINSTANCE g_hInstance = NULL;