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

ShaderGen: Cleanup uid data writing.

This commit is contained in:
NeoBrainX 2013-03-29 21:53:57 +01:00
parent 9eccd56ef0
commit e31c2aa601
3 changed files with 116 additions and 113 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

50
Source/Core/VideoCommon/Src/LightingShaderGen.h
View File

@ -24,18 +24,17 @@
#include "NativeVertexFormat.h" #include "NativeVertexFormat.h"
#include "XFMemory.h" #include "XFMemory.h"
// T.uid_data needs to have a struct named lighting_uid // uid_data needs to have a struct named lighting_uid
template<class T,class UidType> template<class T,class UidDataT>
void GenerateLightShader(T& object, int index, int litchan_index, const char* lightsName, int coloralpha) void GenerateLightShader(T& object, UidDataT& uid_data, int index, int litchan_index, const char* lightsName, int coloralpha)
{ {
#define SetUidField(name, value) if (&object.GetUidData() != NULL) { object.GetUidData().name = value; };
const LitChannel& chan = (litchan_index > 1) ? xfregs.alpha[litchan_index-2] : xfregs.color[litchan_index]; const LitChannel& chan = (litchan_index > 1) ? xfregs.alpha[litchan_index-2] : xfregs.color[litchan_index];
const char* swizzle = "xyzw"; const char* swizzle = "xyzw";
if (coloralpha == 1 ) swizzle = "xyz"; if (coloralpha == 1 ) swizzle = "xyz";
else if (coloralpha == 2 ) swizzle = "w"; else if (coloralpha == 2 ) swizzle = "w";
SetUidField(lit_chans[litchan_index].attnfunc, chan.attnfunc); uid_data.lit_chans[litchan_index].attnfunc = chan.attnfunc;
SetUidField(lit_chans[litchan_index].diffusefunc, chan.diffusefunc); uid_data.lit_chans[litchan_index].diffusefunc = chan.diffusefunc;
if (!(chan.attnfunc & 1)) { if (!(chan.attnfunc & 1)) {
// atten disabled // atten disabled
switch (chan.diffusefunc) { switch (chan.diffusefunc) {
@ -94,8 +93,8 @@ void GenerateLightShader(T& object, int index, int litchan_index, const char* li
// materials name is I_MATERIALS in vs and I_PMATERIALS in ps // materials name is I_MATERIALS in vs and I_PMATERIALS in ps
// inColorName is color in vs and colors_ in ps // inColorName is color in vs and colors_ in ps
// dest is o.colors_ in vs and colors_ in ps // dest is o.colors_ in vs and colors_ in ps
template<class T, class UidType> template<class T, class UidDataT>
void GenerateLightingShader(T& object, int components, const char* materialsName, const char* lightsName, const char* inColorName, const char* dest) void GenerateLightingShader(T& object, UidDataT& uid_data, int components, const char* materialsName, const char* lightsName, const char* inColorName, const char* dest)
{ {
for (unsigned int j = 0; j < xfregs.numChan.numColorChans; j++) for (unsigned int j = 0; j < xfregs.numChan.numColorChans; j++)
{ {
@ -104,7 +103,7 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
object.Write("{\n"); object.Write("{\n");
SetUidField(lit_chans[j].matsource, xfregs.color[j].matsource); uid_data.lit_chans[j].matsource = xfregs.color[j].matsource;
if (color.matsource) {// from vertex if (color.matsource) {// from vertex
if (components & (VB_HAS_COL0 << j)) if (components & (VB_HAS_COL0 << j))
object.Write("mat = %s%d;\n", inColorName, j); object.Write("mat = %s%d;\n", inColorName, j);
@ -116,9 +115,9 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
else // from color else // from color
object.Write("mat = %s[%d];\n", materialsName, j+2); object.Write("mat = %s[%d];\n", materialsName, j+2);
SetUidField(lit_chans[j].enablelighting, xfregs.color[j].enablelighting); uid_data.lit_chans[j].enablelighting = xfregs.color[j].enablelighting;
if (color.enablelighting) { if (color.enablelighting) {
SetUidField(lit_chans[j].ambsource, xfregs.color[j].ambsource); uid_data.lit_chans[j].ambsource = xfregs.color[j].ambsource;
if (color.ambsource) { // from vertex if (color.ambsource) { // from vertex
if (components & (VB_HAS_COL0<<j) ) if (components & (VB_HAS_COL0<<j) )
object.Write("lacc = %s%d;\n", inColorName, j); object.Write("lacc = %s%d;\n", inColorName, j);
@ -136,7 +135,7 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
} }
// check if alpha is different // check if alpha is different
SetUidField(lit_chans[j+2].matsource, xfregs.alpha[j].matsource); uid_data.lit_chans[j+2].matsource = xfregs.alpha[j].matsource;
if (alpha.matsource != color.matsource) { if (alpha.matsource != color.matsource) {
if (alpha.matsource) {// from vertex if (alpha.matsource) {// from vertex
if (components & (VB_HAS_COL0<<j)) if (components & (VB_HAS_COL0<<j))
@ -149,10 +148,10 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
object.Write("mat.w = %s[%d].w;\n", materialsName, j+2); object.Write("mat.w = %s[%d].w;\n", materialsName, j+2);
} }
SetUidField(lit_chans[j+2].enablelighting, xfregs.alpha[j].enablelighting); uid_data.lit_chans[j+2].enablelighting = xfregs.alpha[j].enablelighting;
if (alpha.enablelighting) if (alpha.enablelighting)
{ {
SetUidField(lit_chans[j+2].ambsource, xfregs.alpha[j].ambsource); uid_data.lit_chans[j+2].ambsource = xfregs.alpha[j].ambsource;
if (alpha.ambsource) {// from vertex if (alpha.ambsource) {// from vertex
if (components & (VB_HAS_COL0<<j) ) if (components & (VB_HAS_COL0<<j) )
object.Write("lacc.w = %s%d.w;\n", inColorName, j); object.Write("lacc.w = %s%d.w;\n", inColorName, j);
@ -173,12 +172,12 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
{ {
// both have lighting, test if they use the same lights // both have lighting, test if they use the same lights
int mask = 0; int mask = 0;
SetUidField(lit_chans[j].attnfunc, color.attnfunc); uid_data.lit_chans[j].attnfunc = color.attnfunc;
SetUidField(lit_chans[j+2].attnfunc, alpha.attnfunc); uid_data.lit_chans[j+2].attnfunc = alpha.attnfunc;
SetUidField(lit_chans[j].diffusefunc, color.diffusefunc); uid_data.lit_chans[j].diffusefunc = color.diffusefunc;
SetUidField(lit_chans[j+2].diffusefunc, alpha.diffusefunc); uid_data.lit_chans[j+2].diffusefunc = alpha.diffusefunc;
SetUidField(lit_chans[j].light_mask, color.GetFullLightMask()); uid_data.lit_chans[j].light_mask = color.GetFullLightMask();
SetUidField(lit_chans[j+2].light_mask, alpha.GetFullLightMask()); uid_data.lit_chans[j+2].light_mask = alpha.GetFullLightMask();
if(color.lightparams == alpha.lightparams) if(color.lightparams == alpha.lightparams)
{ {
mask = color.GetFullLightMask() & alpha.GetFullLightMask(); mask = color.GetFullLightMask() & alpha.GetFullLightMask();
@ -188,7 +187,7 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
{ {
if (mask & (1<<i)) if (mask & (1<<i))
{ {
GenerateLightShader<T,UidType>(object, i, j, lightsName, 3); GenerateLightShader<T,UidDataT>(object, uid_data, i, j, lightsName, 3);
} }
} }
} }
@ -198,9 +197,9 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
for (int i = 0; i < 8; ++i) for (int i = 0; i < 8; ++i)
{ {
if (!(mask&(1<<i)) && (color.GetFullLightMask() & (1<<i))) if (!(mask&(1<<i)) && (color.GetFullLightMask() & (1<<i)))
GenerateLightShader<T,UidType>(object, i, j, lightsName, 1); GenerateLightShader<T,UidDataT>(object, uid_data, i, j, lightsName, 1);
if (!(mask&(1<<i)) && (alpha.GetFullLightMask() & (1<<i))) if (!(mask&(1<<i)) && (alpha.GetFullLightMask() & (1<<i)))
GenerateLightShader<T,UidType>(object, i, j+2, lightsName, 2); GenerateLightShader<T,UidDataT>(object, uid_data, i, j+2, lightsName, 2);
} }
} }
else if (color.enablelighting || alpha.enablelighting) else if (color.enablelighting || alpha.enablelighting)
@ -210,17 +209,16 @@ void GenerateLightingShader(T& object, int components, const char* materialsName
const int lit_index = color.enablelighting ? j : (j+2); const int lit_index = color.enablelighting ? j : (j+2);
int coloralpha = color.enablelighting ? 1 : 2; int coloralpha = color.enablelighting ? 1 : 2;
SetUidField(lit_chans[lit_index].light_mask, workingchannel.GetFullLightMask()); uid_data.lit_chans[lit_index].light_mask = workingchannel.GetFullLightMask();
for (int i = 0; i < 8; ++i) for (int i = 0; i < 8; ++i)
{ {
if (workingchannel.GetFullLightMask() & (1<<i)) if (workingchannel.GetFullLightMask() & (1<<i))
GenerateLightShader<T,UidType>(object, i, lit_index, lightsName, coloralpha); GenerateLightShader<T,UidDataT>(object, uid_data, i, lit_index, lightsName, coloralpha);
} }
} }
object.Write("%s%d = mat * saturate(lacc);\n", dest, j); object.Write("%s%d = mat * saturate(lacc);\n", dest, j);
object.Write("}\n"); object.Write("}\n");
} }
} }
#undef SetUidField
#endif // _LIGHTINGSHADERGEN_H_ #endif // _LIGHTINGSHADERGEN_H_

141
Source/Core/VideoCommon/Src/PixelShaderGen.cpp
View File

@ -37,10 +37,10 @@
// output is given by .outreg // output is given by .outreg
// tevtemp is set according to swapmodetables and // tevtemp is set according to swapmodetables and
template<class T> static void WriteStage(char *&p, int n, API_TYPE ApiType); template<class T> static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE ApiType);
template<class T> static void SampleTexture(T& out, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType); template<class T> static void SampleTexture(T& out, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType);
template<class T> static void WriteAlphaTest(T& out, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode, bool per_pixel_depth); template<class T> static void WriteAlphaTest(T& out, pixel_shader_uid_data& uid_data, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode, bool per_pixel_depth);
template<class T> static void WriteFog(T& out); template<class T> static void WriteFog(T& out, pixel_shader_uid_data& uid_data);
static const char *tevKSelTableC[] = // KCSEL static const char *tevKSelTableC[] = // KCSEL
{ {
@ -272,13 +272,15 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
// TODO: Where does this TODO belong again...? // TODO: Where does this TODO belong again...?
// TODO: Can be optimized if using alpha pass // TODO: Can be optimized if using alpha pass
#define SetUidField(name, value) if (&out.GetUidData() != NULL) {out.GetUidData().name = value; }; // Non-uid template parameters will write to the dummy data (=> gets optimized out)
#define OR_UidField(name, value) if (&out.GetUidData() != NULL) {out.GetUidData().name |= value; }; pixel_shader_uid_data dummy_data;
pixel_shader_uid_data& uid_data = (&out.GetUidData() != NULL) ? out.GetUidData() : dummy_data;
out.SetBuffer(text); out.SetBuffer(text);
if (out.GetBuffer() != NULL) if (out.GetBuffer() != NULL)
setlocale(LC_NUMERIC, "C"); // Reset locale for compilation setlocale(LC_NUMERIC, "C"); // Reset locale for compilation
/// text[sizeof(text) - 1] = 0x7C; // canary text[sizeof(text) - 1] = 0x7C; // canary
unsigned int numStages = bpmem.genMode.numtevstages + 1; unsigned int numStages = bpmem.genMode.numtevstages + 1;
unsigned int numTexgen = bpmem.genMode.numtexgens; unsigned int numTexgen = bpmem.genMode.numtexgens;
@ -289,12 +291,12 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
out.Write("//%i TEV stages, %i texgens, XXX IND stages\n", out.Write("//%i TEV stages, %i texgens, XXX IND stages\n",
numStages, numTexgen/*, bpmem.genMode.numindstages*/); numStages, numTexgen/*, bpmem.genMode.numindstages*/);
// SetUidField(components, components); // TODO: Enable once per pixel lighting is implemented again // uid_data.components = components; // TODO: Enable once per pixel lighting is implemented again
SetUidField(dstAlphaMode, dstAlphaMode); uid_data.dstAlphaMode = dstAlphaMode;
SetUidField(genMode.numindstages, bpmem.genMode.numindstages); uid_data.genMode.numindstages = bpmem.genMode.numindstages;
SetUidField(genMode.numtevstages, bpmem.genMode.numtevstages); uid_data.genMode.numtevstages = bpmem.genMode.numtevstages;
SetUidField(genMode.numtexgens, bpmem.genMode.numtexgens); uid_data.genMode.numtexgens = bpmem.genMode.numtexgens;
if (ApiType == API_OPENGL) if (ApiType == API_OPENGL)
{ {
@ -461,7 +463,7 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
if (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting) if (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
{ {
SetUidField(xfregs_numTexGen_numTexGens, xfregs.numTexGen.numTexGens); uid_data.xfregs_numTexGen_numTexGens = xfregs.numTexGen.numTexGens;
if (xfregs.numTexGen.numTexGens < 7) if (xfregs.numTexGen.numTexGens < 7)
{ {
out.Write("\tfloat3 _norm0 = normalize(Normal.xyz);\n\n"); out.Write("\tfloat3 _norm0 = normalize(Normal.xyz);\n\n");
@ -498,7 +500,7 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
for (unsigned int i = 0; i < numTexgen; ++i) for (unsigned int i = 0; i < numTexgen; ++i)
{ {
// optional perspective divides // optional perspective divides
SetUidField(texMtxInfo[i].projection, xfregs.texMtxInfo[i].projection); uid_data.texMtxInfo[i].projection = xfregs.texMtxInfo[i].projection;
if (xfregs.texMtxInfo[i].projection == XF_TEXPROJ_STQ) if (xfregs.texMtxInfo[i].projection == XF_TEXPROJ_STQ)
{ {
out.Write("\tif (uv%d.z != 0.0f)", i); out.Write("\tif (uv%d.z != 0.0f)", i);
@ -520,7 +522,7 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
} }
} }
SetUidField(nIndirectStagesUsed, nIndirectStagesUsed); uid_data.nIndirectStagesUsed = nIndirectStagesUsed;
for (u32 i = 0; i < bpmem.genMode.numindstages; ++i) for (u32 i = 0; i < bpmem.genMode.numindstages; ++i)
{ {
if (nIndirectStagesUsed & (1 << i)) if (nIndirectStagesUsed & (1 << i))
@ -531,23 +533,23 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
/// TODO: Cleanup... /// TODO: Cleanup...
if (i == 0) if (i == 0)
{ {
SetUidField(tevindref.bc0, texcoord); uid_data.tevindref.bc0 = texcoord;
SetUidField(tevindref.bi0, texmap); uid_data.tevindref.bi0 = texmap;
} }
else if (i == 1) else if (i == 1)
{ {
SetUidField(tevindref.bc1, texcoord); uid_data.tevindref.bc1 = texcoord;
SetUidField(tevindref.bi1, texmap); uid_data.tevindref.bi1 = texmap;
} }
else if (i == 2) else if (i == 2)
{ {
SetUidField(tevindref.bc3, texcoord); uid_data.tevindref.bc3 = texcoord;
SetUidField(tevindref.bi2, texmap); uid_data.tevindref.bi2 = texmap;
} }
else else
{ {
SetUidField(tevindref.bc4, texcoord); uid_data.tevindref.bc4 = texcoord;
SetUidField(tevindref.bi4, texmap); uid_data.tevindref.bi4 = texmap;
} }
if (texcoord < numTexgen) if (texcoord < numTexgen)
{ {
@ -576,17 +578,17 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
// Uid fields for BuildSwapModeTable are set in WriteStage // Uid fields for BuildSwapModeTable are set in WriteStage
BuildSwapModeTable(); BuildSwapModeTable();
for (unsigned int i = 0; i < numStages; i++) for (unsigned int i = 0; i < numStages; i++)
WriteStage<T>(out, i, ApiType); // build the equation for this stage WriteStage<T>(out, uid_data, i, ApiType); // build the equation for this stage
if (numStages) if (numStages)
{ {
// The results of the last texenv stage are put onto the screen, // The results of the last texenv stage are put onto the screen,
// regardless of the used destination register // regardless of the used destination register
SetUidField(combiners[numStages-1].colorC.dest, bpmem.combiners[numStages-1].colorC.dest); // TODO: These probably don't need to be set anymore here... uid_data.combiners[numStages-1].colorC.dest = bpmem.combiners[numStages-1].colorC.dest; // TODO: These probably don't need to be set anymore here...
SetUidField(combiners[numStages-1].alphaC.dest, bpmem.combiners[numStages-1].alphaC.dest); uid_data.combiners[numStages-1].alphaC.dest = bpmem.combiners[numStages-1].alphaC.dest;
if(bpmem.combiners[numStages - 1].colorC.dest != 0) if(bpmem.combiners[numStages - 1].colorC.dest != 0)
{ {
/// SetUidField(combiners[numStages-1].colorC.dest, bpmem.combiners[numStages-1].colorC.dest); /// uid_data.combiners[numStages-1].colorC.dest = bpmem.combiners[numStages-1].colorC.dest;
bool retrieveFromAuxRegister = !RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].ColorNeedOverflowControl && RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].AuxStored; bool retrieveFromAuxRegister = !RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].ColorNeedOverflowControl && RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].AuxStored;
out.Write("\tprev.rgb = %s%s;\n", retrieveFromAuxRegister ? "c" : "" , tevCOutputTable[bpmem.combiners[numStages - 1].colorC.dest]); out.Write("\tprev.rgb = %s%s;\n", retrieveFromAuxRegister ? "c" : "" , tevCOutputTable[bpmem.combiners[numStages - 1].colorC.dest]);
RegisterStates[0].ColorNeedOverflowControl = RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].ColorNeedOverflowControl; RegisterStates[0].ColorNeedOverflowControl = RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].ColorNeedOverflowControl;
@ -603,9 +605,9 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
out.Write("\tprev = frac(prev * (255.0f/256.0f)) * (256.0f/255.0f);\n"); out.Write("\tprev = frac(prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");
AlphaTest::TEST_RESULT Pretest = bpmem.alpha_test.TestResult(); AlphaTest::TEST_RESULT Pretest = bpmem.alpha_test.TestResult();
SetUidField(Pretest, Pretest); uid_data.Pretest = Pretest;
if (Pretest == AlphaTest::UNDETERMINED) if (Pretest == AlphaTest::UNDETERMINED)
WriteAlphaTest<T>(out, ApiType, dstAlphaMode, per_pixel_depth); WriteAlphaTest<T>(out, uid_data, ApiType, dstAlphaMode, per_pixel_depth);
// the screen space depth value = far z + (clip z / clip w) * z range // the screen space depth value = far z + (clip z / clip w) * z range
@ -619,9 +621,9 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
} }
// Note: depth textures are disabled if early depth test is enabled // Note: depth textures are disabled if early depth test is enabled
SetUidField(ztex.op, bpmem.ztex2.op); uid_data.ztex.op = bpmem.ztex2.op;
SetUidField(per_pixel_depth, per_pixel_depth); uid_data.per_pixel_depth = per_pixel_depth;
SetUidField(fog.fsel, bpmem.fog.c_proj_fsel.fsel); uid_data.fog.fsel = bpmem.fog.c_proj_fsel.fsel;
// depth texture can safely be ignored if the result won't be written to the depth buffer (early_ztest) and isn't used for fog either // depth texture can safely be ignored if the result won't be written to the depth buffer (early_ztest) and isn't used for fog either
bool skip_ztexture = !per_pixel_depth && !bpmem.fog.c_proj_fsel.fsel; bool skip_ztexture = !per_pixel_depth && !bpmem.fog.c_proj_fsel.fsel;
@ -650,7 +652,7 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
} }
else else
{ {
WriteFog<T>(out); WriteFog<T>(out, uid_data);
out.Write("\tocol0 = prev;\n"); out.Write("\tocol0 = prev;\n");
} }
@ -666,8 +668,9 @@ void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u
} }
out.Write("}\n"); out.Write("}\n");
/// if (text[sizeof(text) - 1] != 0x7C)
/// PanicAlert("PixelShader generator - buffer too small, canary has been eaten!"); if (text[sizeof(text) - 1] != 0x7C)
PanicAlert("PixelShader generator - buffer too small, canary has been eaten!");
if (out.GetBuffer() != NULL) if (out.GetBuffer() != NULL)
setlocale(LC_NUMERIC, ""); // restore locale setlocale(LC_NUMERIC, ""); // restore locale
@ -720,7 +723,7 @@ static const char *TEVCMPAlphaOPTable[16] =
template<class T> template<class T>
static void WriteStage(T& out, int n, API_TYPE ApiType) static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE ApiType)
{ {
int texcoord = bpmem.tevorders[n/2].getTexCoord(n&1); int texcoord = bpmem.tevorders[n/2].getTexCoord(n&1);
bool bHasTexCoord = (u32)texcoord < bpmem.genMode.numtexgens; bool bHasTexCoord = (u32)texcoord < bpmem.genMode.numtexgens;
@ -732,14 +735,14 @@ static void WriteStage(T& out, int n, API_TYPE ApiType)
out.Write("// TEV stage %d\n", n); out.Write("// TEV stage %d\n", n);
OR_UidField(bHasIndStage, bHasIndStage << n); uid_data.bHasIndStage |= bHasIndStage << n;
if (n < 8) { OR_UidField(tevorders_n_texcoord1, texcoord << (3 * n)); } if (n < 8) { uid_data.tevorders_n_texcoord1 |= texcoord << (3 * n); }
else OR_UidField(tevorders_n_texcoord2, texcoord << (3 * n - 24)); else uid_data.tevorders_n_texcoord2 |= texcoord << (3 * n - 24);
if (bHasIndStage) if (bHasIndStage)
{ {
OR_UidField(tevind_n_bs, bpmem.tevind[n].bs << (2*n)); uid_data.tevind_n_bs |= bpmem.tevind[n].bs << (2*n);
OR_UidField(tevind_n_bt, bpmem.tevind[n].bt << (2*n)); uid_data.tevind_n_bt |= bpmem.tevind[n].bt << (2*n);
OR_UidField(tevind_n_fmt, bpmem.tevind[n].fmt << (2*n)); uid_data.tevind_n_fmt |= bpmem.tevind[n].fmt << (2*n);
out.Write("// indirect op\n"); out.Write("// indirect op\n");
// perform the indirect op on the incoming regular coordinates using indtex%d as the offset coords // perform the indirect op on the incoming regular coordinates using indtex%d as the offset coords
@ -754,14 +757,14 @@ static void WriteStage(T& out, int n, API_TYPE ApiType)
out.Write("float3 indtevcrd%d = indtex%d * %s;\n", n, bpmem.tevind[n].bt, tevIndFmtScale[bpmem.tevind[n].fmt]); out.Write("float3 indtevcrd%d = indtex%d * %s;\n", n, bpmem.tevind[n].bt, tevIndFmtScale[bpmem.tevind[n].fmt]);
// bias // bias
if (n < 8) { OR_UidField(tevind_n_bias1, bpmem.tevind[n].bias << (3*n)); } /// XXX: brackets? if (n < 8) { uid_data.tevind_n_bias1 |= bpmem.tevind[n].bias << (3*n); } /// XXX: brackets?
else OR_UidField(tevind_n_bias2, bpmem.tevind[n].bias << (3*n - 24)); else uid_data.tevind_n_bias2 |= bpmem.tevind[n].bias << (3*n - 24);
if (bpmem.tevind[n].bias != ITB_NONE ) if (bpmem.tevind[n].bias != ITB_NONE )
out.Write("indtevcrd%d.%s += %s;\n", n, tevIndBiasField[bpmem.tevind[n].bias], tevIndBiasAdd[bpmem.tevind[n].fmt]); out.Write("indtevcrd%d.%s += %s;\n", n, tevIndBiasField[bpmem.tevind[n].bias], tevIndBiasAdd[bpmem.tevind[n].fmt]);
// multiply by offset matrix and scale // multiply by offset matrix and scale
if (n < 8) { OR_UidField(tevind_n_mid1, bpmem.tevind[n].mid << (4*n)); } /// XXX: brackets? if (n < 8) { uid_data.tevind_n_mid1 |= bpmem.tevind[n].mid << (4*n); } /// XXX: brackets?
else OR_UidField(tevind_n_mid2, bpmem.tevind[n].mid << (4*n - 32)); else uid_data.tevind_n_mid2 |= bpmem.tevind[n].mid << (4*n - 32);
if (bpmem.tevind[n].mid != 0) if (bpmem.tevind[n].mid != 0)
{ {
if (bpmem.tevind[n].mid <= 3) if (bpmem.tevind[n].mid <= 3)
@ -795,12 +798,12 @@ static void WriteStage(T& out, int n, API_TYPE ApiType)
// Wrapping // Wrapping
// --------- // ---------
if (n < 8) { OR_UidField(tevind_n_sw1, bpmem.tevind[n].sw << (3 * n)); } if (n < 8) { uid_data.tevind_n_sw1 |= bpmem.tevind[n].sw << (3 * n); }
else OR_UidField(tevind_n_sw2, bpmem.tevind[n].sw << (3 * n - 24)); else uid_data.tevind_n_sw2 |= bpmem.tevind[n].sw << (3 * n - 24);
if (n < 8) { OR_UidField(tevind_n_tw1, bpmem.tevind[n].tw << (3 * n)); } if (n < 8) { uid_data.tevind_n_tw1 |= bpmem.tevind[n].tw << (3 * n); }
else OR_UidField(tevind_n_tw2, bpmem.tevind[n].tw << (3 * n - 24)); else uid_data.tevind_n_tw2 |= bpmem.tevind[n].tw << (3 * n - 24);
OR_UidField(tevind_n_fb_addprev, bpmem.tevind[n].fb_addprev << n); uid_data.tevind_n_fb_addprev |= bpmem.tevind[n].fb_addprev << n;
// wrap S // wrap S
if (bpmem.tevind[n].sw == ITW_OFF) if (bpmem.tevind[n].sw == ITW_OFF)
@ -827,8 +830,8 @@ static void WriteStage(T& out, int n, API_TYPE ApiType)
TevStageCombiner::ColorCombiner &cc = bpmem.combiners[n].colorC; TevStageCombiner::ColorCombiner &cc = bpmem.combiners[n].colorC;
TevStageCombiner::AlphaCombiner &ac = bpmem.combiners[n].alphaC; TevStageCombiner::AlphaCombiner &ac = bpmem.combiners[n].alphaC;
SetUidField(combiners[n].colorC.hex, cc.hex&0xFFFFFF); uid_data.combiners[n].colorC.hex = cc.hex & 0xFFFFFF;
SetUidField(combiners[n].alphaC.hex, ac.hex&0xFFFFFF); uid_data.combiners[n].alphaC.hex = ac.hex & 0xFFFFFF;
if(cc.a == TEVCOLORARG_RASA || cc.a == TEVCOLORARG_RASC if(cc.a == TEVCOLORARG_RASA || cc.a == TEVCOLORARG_RASC
|| cc.b == TEVCOLORARG_RASA || cc.b == TEVCOLORARG_RASC || cc.b == TEVCOLORARG_RASA || cc.b == TEVCOLORARG_RASC
@ -838,10 +841,10 @@ static void WriteStage(T& out, int n, API_TYPE ApiType)
|| ac.c == TEVALPHAARG_RASA || ac.d == TEVALPHAARG_RASA) || ac.c == TEVALPHAARG_RASA || ac.d == TEVALPHAARG_RASA)
{ {
const int i = bpmem.combiners[n].alphaC.rswap; const int i = bpmem.combiners[n].alphaC.rswap;
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2 ].swap1 << (i*2)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2 ].swap1 << (i*2);
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2+1].swap1 << (i*2 + 1)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2+1].swap1 << (i*2 + 1);
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2 ].swap2 << (i*2 + 16)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2 ].swap2 << (i*2 + 16);
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2+1].swap2 << (i*2 + 17)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2+1].swap2 << (i*2 + 17);
char *rasswap = swapModeTable[bpmem.combiners[n].alphaC.rswap]; char *rasswap = swapModeTable[bpmem.combiners[n].alphaC.rswap];
out.Write("rastemp = %s.%s;\n", tevRasTable[bpmem.tevorders[n / 2].getColorChan(n & 1)], rasswap); out.Write("rastemp = %s.%s;\n", tevRasTable[bpmem.tevorders[n / 2].getColorChan(n & 1)], rasswap);
@ -861,10 +864,10 @@ static void WriteStage(T& out, int n, API_TYPE ApiType)
} }
const int i = bpmem.combiners[n].alphaC.tswap; const int i = bpmem.combiners[n].alphaC.tswap;
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2 ].swap1 << (i*2)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2 ].swap1 << (i*2);
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2+1].swap1 << (i*2 + 1)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2+1].swap1 << (i*2 + 1);
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2 ].swap2 << (i*2 + 16)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2 ].swap2 << (i*2 + 16);
OR_UidField(tevksel_n_swap, bpmem.tevksel[i*2+1].swap2 << (i*2 + 17)); uid_data.tevksel_n_swap |= bpmem.tevksel[i*2+1].swap2 << (i*2 + 17);
char *texswap = swapModeTable[bpmem.combiners[n].alphaC.tswap]; char *texswap = swapModeTable[bpmem.combiners[n].alphaC.tswap];
int texmap = bpmem.tevorders[n/2].getTexMap(n&1); int texmap = bpmem.tevorders[n/2].getTexMap(n&1);
@ -1116,7 +1119,7 @@ static const char *tevAlphaFunclogicTable[] =
}; };
template<class T> template<class T>
static void WriteAlphaTest(T& out, API_TYPE ApiType, DSTALPHA_MODE dstAlphaMode, bool per_pixel_depth) static void WriteAlphaTest(T& out, pixel_shader_uid_data& uid_data, API_TYPE ApiType, DSTALPHA_MODE dstAlphaMode, bool per_pixel_depth)
{ {
static const char *alphaRef[2] = static const char *alphaRef[2] =
{ {
@ -1129,9 +1132,9 @@ static void WriteAlphaTest(T& out, API_TYPE ApiType, DSTALPHA_MODE dstAlphaMode,
// using discard then return works the same in cg and dx9 but not in dx11 // using discard then return works the same in cg and dx9 but not in dx11
out.Write("\tif(!( "); out.Write("\tif(!( ");
SetUidField(alpha_test.comp0, bpmem.alpha_test.comp0); uid_data.alpha_test.comp0 = bpmem.alpha_test.comp0;
SetUidField(alpha_test.logic, bpmem.alpha_test.comp1); uid_data.alpha_test.logic = bpmem.alpha_test.comp1;
SetUidField(alpha_test.logic, bpmem.alpha_test.logic); uid_data.alpha_test.logic = bpmem.alpha_test.logic;
// Lookup the first component from the alpha function table // Lookup the first component from the alpha function table
int compindex = bpmem.alpha_test.comp0; int compindex = bpmem.alpha_test.comp0;
@ -1160,7 +1163,7 @@ static void WriteAlphaTest(T& out, API_TYPE ApiType, DSTALPHA_MODE dstAlphaMode,
// when the alpha test fail. This is not a correct implementation because // when the alpha test fail. This is not a correct implementation because
// even if the depth test fails the fragment could be alpha blended, but // even if the depth test fails the fragment could be alpha blended, but
// we don't have a choice. // we don't have a choice.
SetUidField(alpha_test.use_zcomploc_hack, bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable); uid_data.alpha_test.use_zcomploc_hack = bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable;
if (!(bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable)) if (!(bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable))
{ {
out.Write("\t\tdiscard;\n"); out.Write("\t\tdiscard;\n");
@ -1184,13 +1187,13 @@ static const char *tevFogFuncsTable[] =
}; };
template<class T> template<class T>
static void WriteFog(T& out) static void WriteFog(T& out, pixel_shader_uid_data& uid_data)
{ {
SetUidField(fog.fsel, bpmem.fog.c_proj_fsel.fsel); uid_data.fog.fsel = bpmem.fog.c_proj_fsel.fsel;
if(bpmem.fog.c_proj_fsel.fsel == 0) if(bpmem.fog.c_proj_fsel.fsel == 0)
return; // no Fog return; // no Fog
SetUidField(fog.proj, bpmem.fog.c_proj_fsel.proj); uid_data.fog.proj = bpmem.fog.c_proj_fsel.proj;
out.SetConstantsUsed(C_FOG, C_FOG+1); out.SetConstantsUsed(C_FOG, C_FOG+1);
if (bpmem.fog.c_proj_fsel.proj == 0) if (bpmem.fog.c_proj_fsel.proj == 0)
@ -1209,7 +1212,7 @@ static void WriteFog(T& out)
// x_adjust = sqrt((x-center)^2 + k^2)/k // x_adjust = sqrt((x-center)^2 + k^2)/k
// ze *= x_adjust // ze *= x_adjust
// this is completely theoretical as the real hardware seems to use a table intead of calculating the values. // this is completely theoretical as the real hardware seems to use a table intead of calculating the values.
SetUidField(fog.RangeBaseEnabled, bpmem.fogRange.Base.Enabled); uid_data.fog.RangeBaseEnabled = bpmem.fogRange.Base.Enabled;
if(bpmem.fogRange.Base.Enabled) if(bpmem.fogRange.Base.Enabled)
{ {
out.SetConstantsUsed(C_FOG+2, C_FOG+2); out.SetConstantsUsed(C_FOG+2, C_FOG+2);

38
Source/Core/VideoCommon/Src/VertexShaderGen.cpp
View File

@ -83,14 +83,15 @@ extern const char *WriteLocation(API_TYPE api_type);
template<class T> template<class T>
void GenerateVertexShader(T& out, u32 components, API_TYPE api_type) void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
{ {
#undef SetUidField // Non-uid template parameters will write to the dummy data (=> gets optimized out)
#define SetUidField(name, value) if (&out.GetUidData() != NULL) {out.GetUidData().name = value; }; vertex_shader_uid_data dummy_data;
vertex_shader_uid_data& uid_data = (&out.GetUidData() != NULL) ? out.GetUidData() : dummy_data;
out.SetBuffer(text); out.SetBuffer(text);
if (out.GetBuffer() != NULL) if (out.GetBuffer() != NULL)
setlocale(LC_NUMERIC, "C"); // Reset locale for compilation setlocale(LC_NUMERIC, "C"); // Reset locale for compilation
// text[sizeof(text) - 1] = 0x7C; // canary text[sizeof(text) - 1] = 0x7C; // canary
_assert_(bpmem.genMode.numtexgens == xfregs.numTexGen.numTexGens); _assert_(bpmem.genMode.numtexgens == xfregs.numTexGen.numTexGens);
_assert_(bpmem.genMode.numcolchans == xfregs.numChan.numColorChans); _assert_(bpmem.genMode.numcolchans == xfregs.numChan.numColorChans);
@ -122,8 +123,8 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
GenerateVSOutputStruct(out, components, api_type); GenerateVSOutputStruct(out, components, api_type);
SetUidField(numTexGens, xfregs.numTexGen.numTexGens); uid_data.numTexGens = xfregs.numTexGen.numTexGens;
SetUidField(components, components); uid_data.components = components;
if(api_type == API_OPENGL) if(api_type == API_OPENGL)
{ {
@ -262,7 +263,7 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
"float3 ldir, h;\n" "float3 ldir, h;\n"
"float dist, dist2, attn;\n"); "float dist, dist2, attn;\n");
SetUidField(numColorChans, xfregs.numChan.numColorChans); uid_data.numColorChans = xfregs.numChan.numColorChans;
if (xfregs.numChan.numColorChans == 0) if (xfregs.numChan.numColorChans == 0)
{ {
if (components & VB_HAS_COL0) if (components & VB_HAS_COL0)
@ -272,7 +273,7 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
} }
// TODO: This probably isn't necessary if pixel lighting is enabled. // TODO: This probably isn't necessary if pixel lighting is enabled.
GenerateLightingShader<T,VertexShaderUid>(out, components, I_MATERIALS, I_LIGHTS, "color", "o.colors_"); GenerateLightingShader<T,vertex_shader_uid_data>(out, uid_data, components, I_MATERIALS, I_LIGHTS, "color", "o.colors_");
if (xfregs.numChan.numColorChans < 2) if (xfregs.numChan.numColorChans < 2)
{ {
@ -297,7 +298,7 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
out.Write("{\n"); out.Write("{\n");
out.Write("coord = float4(0.0f, 0.0f, 1.0f, 1.0f);\n"); out.Write("coord = float4(0.0f, 0.0f, 1.0f, 1.0f);\n");
SetUidField(texMtxInfo[i].sourcerow, xfregs.texMtxInfo[i].sourcerow); uid_data.texMtxInfo[i].sourcerow = xfregs.texMtxInfo[i].sourcerow;
switch (texinfo.sourcerow) { switch (texinfo.sourcerow) {
case XF_SRCGEOM_INROW: case XF_SRCGEOM_INROW:
_assert_( texinfo.inputform == XF_TEXINPUT_ABC1 ); _assert_( texinfo.inputform == XF_TEXINPUT_ABC1 );
@ -332,21 +333,21 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
} }
// first transformation // first transformation
SetUidField(texMtxInfo[i].texgentype, xfregs.texMtxInfo[i].texgentype); uid_data.texMtxInfo[i].texgentype = xfregs.texMtxInfo[i].texgentype;
switch (texinfo.texgentype) { switch (texinfo.texgentype) {
case XF_TEXGEN_EMBOSS_MAP: // calculate tex coords into bump map case XF_TEXGEN_EMBOSS_MAP: // calculate tex coords into bump map
if (components & (VB_HAS_NRM1|VB_HAS_NRM2)) { if (components & (VB_HAS_NRM1|VB_HAS_NRM2)) {
// transform the light dir into tangent space // transform the light dir into tangent space
SetUidField(texMtxInfo[i].embosslightshift, xfregs.texMtxInfo[i].embosslightshift); uid_data.texMtxInfo[i].embosslightshift = xfregs.texMtxInfo[i].embosslightshift;
SetUidField(texMtxInfo[i].embosssourceshift, xfregs.texMtxInfo[i].embosssourceshift); uid_data.texMtxInfo[i].embosssourceshift = xfregs.texMtxInfo[i].embosssourceshift;
out.Write("ldir = normalize(" I_LIGHTS"[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift); out.Write("ldir = normalize(" I_LIGHTS"[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift);
out.Write("o.tex%d.xyz = o.tex%d.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift); out.Write("o.tex%d.xyz = o.tex%d.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift);
} }
else else
{ {
_assert_(0); // should have normals _assert_(0); // should have normals
SetUidField(texMtxInfo[i].embosssourceshift, xfregs.texMtxInfo[i].embosssourceshift); uid_data.texMtxInfo[i].embosssourceshift = xfregs.texMtxInfo[i].embosssourceshift;
out.Write("o.tex%d.xyz = o.tex%d.xyz;\n", i, texinfo.embosssourceshift); out.Write("o.tex%d.xyz = o.tex%d.xyz;\n", i, texinfo.embosssourceshift);
} }
@ -361,7 +362,7 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
break; break;
case XF_TEXGEN_REGULAR: case XF_TEXGEN_REGULAR:
default: default:
SetUidField(texMtxInfo[i].projection, xfregs.texMtxInfo[i].projection); uid_data.texMtxInfo[i].projection = xfregs.texMtxInfo[i].projection;
if (components & (VB_HAS_TEXMTXIDX0<<i)) if (components & (VB_HAS_TEXMTXIDX0<<i))
{ {
out.Write("int tmp = int(tex%d.z);\n", i); out.Write("int tmp = int(tex%d.z);\n", i);
@ -380,11 +381,11 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
break; break;
} }
SetUidField(dualTexTrans.enabled, xfregs.dualTexTrans.enabled); uid_data.dualTexTrans.enabled = xfregs.dualTexTrans.enabled;
if (xfregs.dualTexTrans.enabled && texinfo.texgentype == XF_TEXGEN_REGULAR) { // only works for regular tex gen types? if (xfregs.dualTexTrans.enabled && texinfo.texgentype == XF_TEXGEN_REGULAR) { // only works for regular tex gen types?
const PostMtxInfo& postInfo = xfregs.postMtxInfo[i]; const PostMtxInfo& postInfo = xfregs.postMtxInfo[i];
SetUidField(postMtxInfo[i].index, xfregs.postMtxInfo[i].index); uid_data.postMtxInfo[i].index = xfregs.postMtxInfo[i].index;
int postidx = postInfo.index; int postidx = postInfo.index;
out.Write("float4 P0 = " I_POSTTRANSFORMMATRICES"[%d];\n" out.Write("float4 P0 = " I_POSTTRANSFORMMATRICES"[%d];\n"
"float4 P1 = " I_POSTTRANSFORMMATRICES"[%d];\n" "float4 P1 = " I_POSTTRANSFORMMATRICES"[%d];\n"
@ -401,7 +402,7 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
} }
else else
{ {
SetUidField(postMtxInfo[i].normalize, xfregs.postMtxInfo[i].normalize); uid_data.postMtxInfo[i].normalize = xfregs.postMtxInfo[i].normalize;
if (postInfo.normalize) if (postInfo.normalize)
out.Write("o.tex%d.xyz = normalize(o.tex%d.xyz);\n", i, i); out.Write("o.tex%d.xyz = normalize(o.tex%d.xyz);\n", i, i);
@ -517,8 +518,9 @@ void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
else else
out.Write("return o;\n}\n"); out.Write("return o;\n}\n");
/// if (text[sizeof(text) - 1] != 0x7C) if (text[sizeof(text) - 1] != 0x7C)
/// PanicAlert("VertexShader generator - buffer too small, canary has been eaten!"); PanicAlert("VertexShader generator - buffer too small, canary has been eaten!");
if (out.GetBuffer() != NULL) if (out.GetBuffer() != NULL)
setlocale(LC_NUMERIC, ""); // restore locale setlocale(LC_NUMERIC, ""); // restore locale
} }