second try, a compilable :) fix for issue 3976 and some optimizations to color loading that will not mess with colors.

git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@6917 8ced0084-cf51-0410-be5f-012b33b47a6e
This commit is contained in:
Rodolfo Osvaldo Bogado 2011-01-25 22:30:15 +00:00
parent 994913ce07
commit 4c704be0ed
3 changed files with 44 additions and 46 deletions

View File

@ -26,7 +26,7 @@
// shader cache for every revision, graphics-related or not, which is simply annoying. // shader cache for every revision, graphics-related or not, which is simply annoying.
enum enum
{ {
LINEAR_DISKCACHE_VER = 6908 LINEAR_DISKCACHE_VER = 6917
}; };
// On disk format: // On disk format:

View File

@ -163,7 +163,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode)
static void WriteStage(char *&p, int n, API_TYPE ApiType); static void WriteStage(char *&p, int n, API_TYPE ApiType);
static void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType); static void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType);
// static void WriteAlphaCompare(char *&p, int num, int comp); // static void WriteAlphaCompare(char *&p, int num, int comp);
static bool WriteAlphaTest(char *&p, API_TYPE ApiType); static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode);
static void WriteFog(char *&p); static void WriteFog(char *&p);
static int AlphaPreTest(); static int AlphaPreTest();
@ -796,7 +796,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
if(RegisterStates[0].AlphaNeedOverflowControl || RegisterStates[0].ColorNeedOverflowControl) if(RegisterStates[0].AlphaNeedOverflowControl || RegisterStates[0].ColorNeedOverflowControl)
WRITE(p, "prev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n"); WRITE(p, "prev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");
if (!WriteAlphaTest(p, ApiType)) if (!WriteAlphaTest(p, ApiType, dstAlphaMode))
{ {
// alpha test will always fail, so restart the shader and just make it an empty function // alpha test will always fail, so restart the shader and just make it an empty function
p = pmainstart; p = pmainstart;
@ -1290,7 +1290,7 @@ static int AlphaPreTest()
} }
static bool WriteAlphaTest(char *&p, API_TYPE ApiType) static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode)
{ {
int Pretest = AlphaPreTest(); int Pretest = AlphaPreTest();
@ -1309,7 +1309,7 @@ static bool WriteAlphaTest(char *&p, API_TYPE ApiType)
compindex = bpmem.alphaFunc.comp1 % 8; compindex = bpmem.alphaFunc.comp1 % 8;
WRITE(p, tevAlphaFuncsTable[compindex],alphaRef[1]);//lookup the second component from the alpha function table WRITE(p, tevAlphaFuncsTable[compindex],alphaRef[1]);//lookup the second component from the alpha function table
WRITE(p, ")){ocol0 = 0;%sdiscard;%s}\n",DepthTextureEnable ? "depth = 1.f;" : "",(ApiType != API_D3D11)? "return;" : ""); WRITE(p, ")){ocol0 = 0;%s%sdiscard;%s}\n",dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND ? "ocol1 = 0;" : "",DepthTextureEnable ? "depth = 1.f;" : "",(ApiType != API_D3D11)? "return;" : "");
return true; return true;
} }

View File

@ -29,10 +29,12 @@
#define GSHIFT 8 #define GSHIFT 8
#define BSHIFT 16 #define BSHIFT 16
#define ASHIFT 24 #define ASHIFT 24
#define AMASK 24
extern int colIndex; extern int colIndex;
extern int colElements[2]; extern int colElements[2];
__forceinline void _SetCol(u32 val) __forceinline void _SetCol(u32 val)
{ {
*(u32*)VertexManager::s_pCurBufferPointer = val; *(u32*)VertexManager::s_pCurBufferPointer = val;
@ -40,36 +42,45 @@ __forceinline void _SetCol(u32 val)
colIndex++; colIndex++;
} }
//color comes in format BARG in 16 bits
//BARG -> AABBGGRR
__forceinline void _SetCol4444(u16 val) __forceinline void _SetCol4444(u16 val)
{ {
u32 col = Convert4To8(val & 0xF) << ASHIFT; u32 col = (val & 0xF0); // col = 000000R0;
col |= Convert4To8((val >> 12) & 0xF) << RSHIFT; col |= (val & 0xF ) << 12; // col |= 0000G000;
col |= Convert4To8((val >> 8) & 0xF) << GSHIFT; col |= (((u32)val) & 0xF000) << 8; // col |= 00B00000;
col |= Convert4To8((val >> 4) & 0xF) << BSHIFT; col |= (((u32)val) & 0x0F00) << 20; // col |= A0000000;
col |= col >> 4; // col = A0B0G0R0 | 0A0B0G0R;
_SetCol(col); _SetCol(col);
} }
//color comes in format RGBA
//RRRRRRGG GGGGBBBB BBAAAAAA
__forceinline void _SetCol6666(u32 val) __forceinline void _SetCol6666(u32 val)
{ {
u32 col = Convert6To8((val >> 18) & 0x3F) << RSHIFT; u32 col = (val >> 16) & 0xFC;
col |= Convert6To8((val >> 12) & 0x3F) << GSHIFT; col |= (val >> 2) & 0xFC00;
col |= Convert6To8((val >> 6) & 0x3F) << BSHIFT; col |= (val << 12) & 0xFC0000;
col |= Convert6To8(val & 0x3F) << ASHIFT; col |= (val << 26) & 0xFC000000;
col |= (col >> 6) & 0x03030303;
_SetCol(col); _SetCol(col);
} }
//color comes in RGB
//RRRRRGGG GGGBBBBB
__forceinline void _SetCol565(u16 val) __forceinline void _SetCol565(u16 val)
{ {
u32 col = Convert5To8((val >> 11) & 0x1F) << RSHIFT; u32 col = (val >> 8) & 0xF8;
col |= Convert6To8((val >> 5) & 0x3F) << GSHIFT; col |= (val << 3) & 0xFC00;
col |= Convert5To8(val & 0x1F) << BSHIFT; col |=(((u32)val) << 19) & 0xF80000;
_SetCol(col | (0xFF << ASHIFT)); col |= (col >> 5) & 0x070007;
col |= (col >> 6) & 0x000300;
_SetCol(col | AMASK);
} }
__forceinline u32 _Read24(const u8 *addr) __forceinline u32 _Read24(const u8 *addr)
{ {
return addr[0] | (addr[1] << 8) | (addr[2] << 16) | 0xFF000000; return (*(const u32 *)addr) | AMASK;
} }
__forceinline u32 _Read32(const u8 *addr) __forceinline u32 _Read32(const u8 *addr)
@ -78,22 +89,15 @@ __forceinline u32 _Read32(const u8 *addr)
} }
void LOADERDECL Color_ReadDirect_24b_888() void LOADERDECL Color_ReadDirect_24b_888()
{ {
u32 col = DataReadU8() << RSHIFT; _SetCol(_Read24(DataGetPosition()));
col |= DataReadU8() << GSHIFT; DataSkip(3);
col |= DataReadU8() << BSHIFT;
_SetCol(col | (0xFF << ASHIFT));
} }
void LOADERDECL Color_ReadDirect_32b_888x(){ void LOADERDECL Color_ReadDirect_32b_888x(){
u32 col = DataReadU8() << RSHIFT; _SetCol(_Read24(DataGetPosition()));
col |= DataReadU8() << GSHIFT; DataSkip(4);
col |= DataReadU8() << BSHIFT;
_SetCol(col | (0xFF << ASHIFT));
DataReadU8();
} }
void LOADERDECL Color_ReadDirect_16b_565() void LOADERDECL Color_ReadDirect_16b_565()
{ {
@ -101,16 +105,14 @@ void LOADERDECL Color_ReadDirect_16b_565()
} }
void LOADERDECL Color_ReadDirect_16b_4444() void LOADERDECL Color_ReadDirect_16b_4444()
{ {
_SetCol4444(DataReadU16()); _SetCol4444(*(u16*)DataGetPosition());
DataSkip(2);
} }
void LOADERDECL Color_ReadDirect_24b_6666() void LOADERDECL Color_ReadDirect_24b_6666()
{ {
u32 val = DataReadU8() << 16; _SetCol6666(Common::swap32(DataGetPosition() - 1));
val |= DataReadU8() << 8; DataSkip(3);
val |= DataReadU8();
_SetCol6666(val);
} }
// F|RES: i am not 100 percent sure, but the colElements seems to be important for rendering only // F|RES: i am not 100 percent sure, but the colElements seems to be important for rendering only
// at least it fixes mario party 4 // at least it fixes mario party 4
// //
@ -153,14 +155,14 @@ void LOADERDECL Color_ReadIndex8_32b_888x()
void LOADERDECL Color_ReadIndex8_16b_4444() void LOADERDECL Color_ReadIndex8_16b_4444()
{ {
u8 Index = DataReadU8(); u8 Index = DataReadU8();
u16 val = Common::swap16(*(const u16 *)(cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]))); u16 val = *(const u16 *)(cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]));
_SetCol4444(val); _SetCol4444(val);
} }
void LOADERDECL Color_ReadIndex8_24b_6666() void LOADERDECL Color_ReadIndex8_24b_6666()
{ {
u8 Index = DataReadU8(); u8 Index = DataReadU8();
const u8* pData = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]); const u8* pData = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]) - 1;
u32 val = pData[2] | (pData[1] << 8) | (pData[0] << 16); u32 val = Common::swap32(pData);
_SetCol6666(val); _SetCol6666(val);
} }
void LOADERDECL Color_ReadIndex8_32b_8888() void LOADERDECL Color_ReadIndex8_32b_8888()
@ -169,9 +171,6 @@ void LOADERDECL Color_ReadIndex8_32b_8888()
const u8 *iAddress = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]); const u8 *iAddress = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]);
_SetCol(_Read32(iAddress)); _SetCol(_Read32(iAddress));
} }
void LOADERDECL Color_ReadIndex16_16b_565() void LOADERDECL Color_ReadIndex16_16b_565()
{ {
u16 Index = DataReadU16(); u16 Index = DataReadU16();
@ -193,14 +192,14 @@ void LOADERDECL Color_ReadIndex16_32b_888x()
void LOADERDECL Color_ReadIndex16_16b_4444() void LOADERDECL Color_ReadIndex16_16b_4444()
{ {
u16 Index = DataReadU16(); u16 Index = DataReadU16();
u16 val = Common::swap16(*(const u16 *)(cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]))); u16 val = *(const u16 *)(cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]));
_SetCol4444(val); _SetCol4444(val);
} }
void LOADERDECL Color_ReadIndex16_24b_6666() void LOADERDECL Color_ReadIndex16_24b_6666()
{ {
u16 Index = DataReadU16(); u16 Index = DataReadU16();
const u8 *pData = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]); const u8 *pData = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]) - 1;
u32 val = pData[2] | (pData[1] << 8) | (pData[0] << 16); u32 val = Common::swap32(pData);
_SetCol6666(val); _SetCol6666(val);
} }
void LOADERDECL Color_ReadIndex16_32b_8888() void LOADERDECL Color_ReadIndex16_32b_8888()
@ -209,5 +208,4 @@ void LOADERDECL Color_ReadIndex16_32b_8888()
const u8 *iAddress = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]); const u8 *iAddress = cached_arraybases[ARRAY_COLOR+colIndex] + (Index * arraystrides[ARRAY_COLOR+colIndex]);
_SetCol(_Read32(iAddress)); _SetCol(_Read32(iAddress));
} }
#endif #endif