piwalker/dolphin
1
0
Fork 0
mirror of https://github.com/dolphin-emu/dolphin.git synced 2025-07-23 22:29:39 -06:00
Code Issues Packages Projects Releases Wiki Activity

Roll back r3833 for render targets but keep it for static textures. Most mirroring issues are OK and we shouldn't see any slowdown.

Browse Source 
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@3844 8ced0084-cf51-0410-be5f-012b33b47a6e
This commit is contained in:
hrydgard
2009-07-19 08:17:41 +00:00
parent ad440b9e47
commit fd70f99f04
11 changed files with 220 additions and 90 deletions
Download Patch File Download Diff File Expand all files Collapse all files

81
Source/Core/VideoCommon/Src/PixelShaderGen.cpp
View File

@ -28,7 +28,7 @@
// a unique identifier, basically containing all the bits. Yup, it's a lot ....
// It would likely be a lot more efficient to build this incrementally as the attributes
// are set...
void GetPixelShaderId(PIXELSHADERUID &uid, u32 dstAlphaEnable)
void GetPixelShaderId(PIXELSHADERUID &uid, u32 s_texturemask, u32 dstAlphaEnable)
{
u32 projtexcoords = 0;
for (u32 i = 0; i < (u32)bpmem.genMode.numtevstages + 1; i++) {
@ -51,7 +51,7 @@ void GetPixelShaderId(PIXELSHADERUID &uid, u32 dstAlphaEnable)
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] = 0;
uid.values[2] = s_texturemask;
uid.values[3] = (u32)bpmem.fog.c_proj_fsel.fsel |
((u32)bpmem.fog.c_proj_fsel.proj << 3);
@ -130,8 +130,8 @@ void GetPixelShaderId(PIXELSHADERUID &uid, u32 dstAlphaEnable)
// output is given by .outreg
// tevtemp is set according to swapmodetables and
static void WriteStage(char *&p, int n);
static void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap);
static void WriteStage(char *&p, int n, u32 texture_mask);
static void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, u32 texture_mask);
static void WriteAlphaCompare(char *&p, int num, int comp);
static bool WriteAlphaTest(char *&p, bool HLSL);
static void WriteFog(char *&p);
@ -368,7 +368,7 @@ static void BuildSwapModeTable()
}
}
const char *GeneratePixelShader(bool dstAlphaEnable, bool HLSL)
const char *GeneratePixelShader(u32 texture_mask, bool dstAlphaEnable, bool HLSL)
{
text[sizeof(text) - 1] = 0x7C; // canary
DVSTARTPROFILE();
@ -391,13 +391,30 @@ const char *GeneratePixelShader(bool dstAlphaEnable, bool HLSL)
}
}
WRITE(p, "uniform sampler2D ");
bool bfirst = true;
for (int i = 0; i < 8; ++i) {
WRITE(p, "%s samp%d : register(s%d)", bfirst ? "" : ",",i, i);
bfirst = false;
// Declare samplers
if (texture_mask) {
WRITE(p, "uniform samplerRECT ");
bool bfirst = true;
for (int i = 0; i < 8; ++i) {
if (texture_mask & (1<<i)) {
WRITE(p, "%s samp%d : register(s%d)", bfirst?"":",", i, i);
bfirst = false;
}
}
WRITE(p, ";\n");
}
if (texture_mask != 0xff) {
WRITE(p, "uniform sampler2D ");
bool bfirst = true;
for (int i = 0; i < 8; ++i) {
if (!(texture_mask & (1<<i))) {
WRITE(p, "%s samp%d : register(s%d)", bfirst?"":",",i, i);
bfirst = false;
}
}
WRITE(p, ";\n");
}
WRITE(p, ";\n");
WRITE(p, "\n");
@ -459,7 +476,7 @@ const char *GeneratePixelShader(bool dstAlphaEnable, bool HLSL)
char buffer[32];
sprintf(buffer, "float3 indtex%d", i);
SampleTexture(p, buffer, "tempcoord", "abg", bpmem.tevindref.getTexMap(i));
SampleTexture(p, buffer, "tempcoord", "abg", bpmem.tevindref.getTexMap(i), texture_mask);
}
}
@ -469,7 +486,7 @@ const char *GeneratePixelShader(bool dstAlphaEnable, bool HLSL)
}
for (int i = 0; i < numStages; i++)
WriteStage(p, i); //build the equation for this stage
WriteStage(p, i, texture_mask); //build the equation for this stage
if (numTexgen >= 7) {
WRITE(p, "float4 clipPos = float4(uv0.w, uv1.w, uv2.w, uv3.w);\n");
@ -512,7 +529,7 @@ const char *GeneratePixelShader(bool dstAlphaEnable, bool HLSL)
return text;
}
static void WriteStage(char *&p, int n)
static void WriteStage(char *&p, int n, u32 texture_mask)
{
char *rasswap = swapModeTable[bpmem.combiners[n].alphaC.rswap];
char *texswap = swapModeTable[bpmem.combiners[n].alphaC.tswap];
@ -612,7 +629,7 @@ static void WriteStage(char *&p, int n)
}
}
WRITE(p, "rastemp=%s.%s;\n", tevRasTable[bpmem.tevorders[n / 2].getColorChan(n & 1)], rasswap);
WRITE(p, "rastemp=%s.%s;\n", tevRasTable[bpmem.tevorders[n / 2].getColorChan(n & 1)],rasswap);
if (bpmem.tevorders[n/2].getEnable(n&1)) {
int texmap = bpmem.tevorders[n/2].getTexMap(n&1);
@ -625,7 +642,7 @@ static void WriteStage(char *&p, int n)
}
}
SampleTexture(p, "textemp", "tevcoord", texswap, texmap);
SampleTexture(p, "textemp", "tevcoord", texswap, texmap, texture_mask);
}
else
WRITE(p, "textemp=float4(1,1,1,1);\n");
@ -732,9 +749,37 @@ static void WriteStage(char *&p, int n)
WRITE(p, "\n");
}
void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap)
void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, u32 texture_mask)
{
WRITE(p, "%s=tex2D(samp%d,%s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap);
if (texture_mask & (1<<texmap)) {
// non pow 2
bool bwraps = (texture_mask & (0x100<<texmap)) ? true : false;
bool bwrapt = (texture_mask & (0x10000<<texmap)) ? true : false;
if (bwraps || bwrapt) {
if (bwraps) {
WRITE(p, "tempcoord.x = fmod(%s.x, "I_TEXDIMS"[%d].x);\n", texcoords, texmap);
}
else {
WRITE(p, "tempcoord.x = %s.x;\n", texcoords);
}
if (bwrapt) {
WRITE(p, "tempcoord.y = fmod(%s.y, "I_TEXDIMS"[%d].y);\n", texcoords, texmap);
}
else {
WRITE(p, "tempcoord.y = %s.y;\n", texcoords);
}
WRITE(p, "%s=texRECT(samp%d,tempcoord.xy).%s;\n", destination, texmap, texswap);
}
else {
WRITE(p, "%s=texRECT(samp%d,%s.xy).%s;\n", destination, texmap, texcoords, texswap);
}
}
else {
WRITE(p, "%s=tex2D(samp%d,%s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap);
}
}
static void WriteAlphaCompare(char *&p, int num, int comp)

4
Source/Core/VideoCommon/Src/PixelShaderGen.h
View File

@ -92,7 +92,7 @@ public:
}
};
const char *GeneratePixelShader(bool dstAlphaEnable, bool HLSL = false);
void GetPixelShaderId(PIXELSHADERUID &, u32 dstAlphaEnable);
const char *GeneratePixelShader(u32 texture_mask, bool dstAlphaEnable, bool HLSL = false);
void GetPixelShaderId(PIXELSHADERUID &, u32 s_texturemask, u32 dstAlphaEnable);
#endif

54
Source/Core/VideoCommon/Src/PixelShaderManager.cpp
View File

@ -40,6 +40,11 @@ static u32 lastAlpha = 0;
static u32 lastTexDims[8]={0}; // width | height << 16 | wrap_s << 28 | wrap_t << 30
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;
void PixelShaderManager::Init()
{
s_nColorsChanged[0] = s_nColorsChanged[1] = 0;
@ -203,15 +208,26 @@ void PixelShaderManager::SetConstants()
void PixelShaderManager::SetPSTextureDims(int texid)
{
// texdims.xy are reciprocals of the real texture dimensions
// texdims.zw are the scaled dimensions
// non pow 2 textures - texdims.xy are the real texture dimensions used for wrapping
// pow 2 textures - texdims.xy are reciprocals of the real texture dimensions
// both - texdims.zw are the scaled dimensions
float fdims[4];
TCoordInfo& tc = bpmem.texcoords[texid];
fdims[0] = 1.0f / (float)(lastTexDims[texid] & 0xffff);
fdims[1] = 1.0f / (float)((lastTexDims[texid] >> 16) & 0xfff);
fdims[2] = (float)(tc.s.scale_minus_1 + 1) * lastCustomTexScale[texid][0];
fdims[3] = (float)(tc.t.scale_minus_1 + 1) * lastCustomTexScale[texid][1];
if (s_texturemask & (1 << texid))
{
TCoordInfo& tc = bpmem.texcoords[texid];
fdims[0] = (float)(lastTexDims[texid] & 0xffff);
fdims[1] = (float)((lastTexDims[texid] >> 16) & 0xfff);
fdims[2] = (float)(tc.s.scale_minus_1 + 1)*lastCustomTexScale[texid][0];
fdims[3] = (float)(tc.t.scale_minus_1 + 1)*lastCustomTexScale[texid][1];
}
else
{
TCoordInfo& tc = bpmem.texcoords[texid];
fdims[0] = 1.0f / (float)(lastTexDims[texid] & 0xffff);
fdims[1] = 1.0f / (float)((lastTexDims[texid] >> 16) & 0xfff);
fdims[2] = (float)(tc.s.scale_minus_1 + 1) * lastCustomTexScale[texid][0];
fdims[3] = (float)(tc.t.scale_minus_1 + 1) * lastCustomTexScale[texid][1];
}
PRIM_LOG("texdims%d: %f %f %f %f\n", texid, fdims[0], fdims[1], fdims[2], fdims[3]);
SetPSConstant4fv(C_TEXDIMS + texid, fdims);
@ -312,6 +328,23 @@ void PixelShaderManager::SetZTextureTypeChanged()
s_bZTextureTypeChanged = true;
}
void PixelShaderManager::SetTexturesUsed(u32 nonpow2tex)
{
if (s_texturemask != nonpow2tex)
{
for (int i = 0; i < 8; ++i)
{
if (nonpow2tex & (0x10101 << i))
{
// this check was previously implicit, but should it be here?
if (s_nTexDimsChanged )
s_nTexDimsChanged |= 1 << i;
}
}
s_texturemask = nonpow2tex;
}
}
void PixelShaderManager::SetTexCoordChanged(u8 texmapid)
{
s_nTexDimsChanged |= 1 << texmapid;
@ -335,3 +368,8 @@ void PixelShaderManager::SetColorMatrix(const float* pmatrix, const float* pfCon
SetPSConstant4fv(C_COLORMATRIX+3, pmatrix+12);
SetPSConstant4fv(C_COLORMATRIX+4, pfConstAdd);
}
u32 PixelShaderManager::GetTextureMask()
{
return s_texturemask;
}

2
Source/Core/VideoCommon/Src/PixelShaderManager.h
View File

@ -46,10 +46,12 @@ public:
static void SetTevKSelChanged(int id);
static void SetZTextureTypeChanged();
static void SetIndTexScaleChanged(u8 stagemask);
static void SetTexturesUsed(u32 nonpow2tex);
static void SetTexCoordChanged(u8 texmapid);
static void SetFogColorChanged();
static void SetFogParamChanged();
static void SetColorMatrix(const float* pmatrix, const float* pfConstAdd);
static u32 GetTextureMask();
};

9
Source/Core/VideoCommon/Src/VertexShaderGen.cpp
View File

@ -410,7 +410,7 @@ const char *GenerateVertexShader(u32 components)
break;
}
if (xfregs.bEnableDualTexTransform && texinfo.texgentype == XF_TEXGEN_REGULAR) { // only works for regular tex gen types?
if(xfregs.bEnableDualTexTransform && texinfo.texgentype == XF_TEXGEN_REGULAR) { // only works for regular tex gen types?
if (xfregs.texcoords[i].postmtxinfo.normalize)
WRITE(p, "o.tex%d.xyz = normalize(o.tex%d.xyz);\n", i, i);
@ -470,8 +470,7 @@ char* GenerateLightShader(char* p, int index, const LitChannel& chan, const char
if (coloralpha == 1 ) swizzle = "xyz";
else if (coloralpha == 2 ) swizzle = "w";
if (!(chan.attnfunc & 1))
{
if (!(chan.attnfunc & 1)) {
// atten disabled
switch (chan.diffusefunc) {
case LIGHTDIF_NONE:
@ -486,9 +485,7 @@ char* GenerateLightShader(char* p, int index, const LitChannel& chan, const char
default: _assert_(0);
}
}
else
{
// spec and spot
else { // spec and spot
WRITE(p, "ldir = "I_LIGHTS".lights[%d].pos.xyz - pos.xyz;\n", index);
if (chan.attnfunc == 3) { // spot