Skip to content

S
swiftshader
Commit c2534f4b by Nicolas Capens
Options

Refactor sampler LOD calculation method into an enum.

Change-Id: I0beed96cd68608ce07aec0b11f14d6f61e67e53d Reviewed-on: https://swiftshader-review.googlesource.com/5034Tested-by: 's avatarNicolas Capens <capn@google.com> Reviewed-by: 's avatarAlexis Hétu <sugoi@google.com> Reviewed-by: 's avatarNicolas Capens <capn@google.com>
parent 8776c531
Showing with 171 additions and 177 deletions
src/Shader/PixelPipeline.cpp
...@@ -94,9 +94,9 @@ namespace sw ...@@ -94,9 +94,9 @@ namespace sw
Vector4s s1; Vector4s s1;
Vector4s s2; Vector4s s2;
if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegisterS(src0); if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegister(src0);
if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegisterS(src1); if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegister(src1);
if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegisterS(src2); if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegister(src2);
Float4 x = version < 0x0104 ? v[2 + dst.index].x : v[2 + src0.index].x; Float4 x = version < 0x0104 ? v[2 + dst.index].x : v[2 + src0.index].x;
Float4 y = version < 0x0104 ? v[2 + dst.index].y : v[2 + src0.index].y; Float4 y = version < 0x0104 ? v[2 + dst.index].y : v[2 + src0.index].y;
...@@ -1221,25 +1221,20 @@ namespace sw ...@@ -1221,25 +1221,20 @@ namespace sw
sampleTexture(c, stage, x, y, z, w, project); sampleTexture(c, stage, x, y, z, w, project);
} }
void PixelPipeline::sampleTexture(Vector4s &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project, bool bias) void PixelPipeline::sampleTexture(Vector4s &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project)
{ {
#if PERF_PROFILE
Long texTime = Ticks();
#endif
Vector4f dsx; Vector4f dsx;
Vector4f dsy; Vector4f dsy;
sampleTexture(c, stage, u, v, w, q, dsx, dsy, project, bias, false);
}
void PixelPipeline::sampleTexture(Vector4s &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, bool bias, bool gradients, bool lodProvided)
{
#if PERF_PROFILE
Long texTime = Ticks();
#endif
Pointer<Byte> texture = data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture); Pointer<Byte> texture = data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture);
if(!project) if(!project)
{ {
sampler[stage]->sampleTexture(texture, c, u, v, w, q, dsx, dsy, bias, gradients, lodProvided); sampler[stage]->sampleTexture(texture, c, u, v, w, q, dsx, dsy);
} }
else else
{ {
...@@ -1249,12 +1244,12 @@ namespace sw ...@@ -1249,12 +1244,12 @@ namespace sw
Float4 v_q = v * rq; Float4 v_q = v * rq;
Float4 w_q = w * rq; Float4 w_q = w * rq;
sampler[stage]->sampleTexture(texture, c, u_q, v_q, w_q, q, dsx, dsy, bias, gradients, lodProvided); sampler[stage]->sampleTexture(texture, c, u_q, v_q, w_q, q, dsx, dsy);
} }
#if PERF_PROFILE #if PERF_PROFILE
cycles[PERF_TEX] += Ticks() - texTime; cycles[PERF_TEX] += Ticks() - texTime;
#endif #endif
} }
Short4 PixelPipeline::convertFixed12(RValue<Float4> cf) Short4 PixelPipeline::convertFixed12(RValue<Float4> cf)
...@@ -1317,7 +1312,7 @@ namespace sw ...@@ -1317,7 +1312,7 @@ namespace sw
} }
} }
Vector4s PixelPipeline::fetchRegisterS(const Src &src) Vector4s PixelPipeline::fetchRegister(const Src &src)
{ {
Vector4s *reg; Vector4s *reg;
int i = src.index; int i = src.index;
......
src/Shader/PixelPipeline.hpp
...@@ -57,8 +57,7 @@ namespace sw ...@@ -57,8 +57,7 @@ namespace sw
void specularPixel(Vector4s &current, Vector4s &specular); void specularPixel(Vector4s &current, Vector4s &specular);
void sampleTexture(Vector4s &c, int coordinates, int sampler, bool project = false); void sampleTexture(Vector4s &c, int coordinates, int sampler, bool project = false);
void sampleTexture(Vector4s &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project = false, bool bias = false); void sampleTexture(Vector4s &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project = false);
void sampleTexture(Vector4s &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project = false, bool bias = false, bool gradients = false, bool lodProvided = false);
Short4 convertFixed12(RValue<Float4> cf); Short4 convertFixed12(RValue<Float4> cf);
void convertFixed12(Vector4s &cs, Vector4f &cf); void convertFixed12(Vector4s &cs, Vector4f &cf);
...@@ -66,7 +65,7 @@ namespace sw ...@@ -66,7 +65,7 @@ namespace sw
void convertSigned12(Vector4f &cf, Vector4s &cs); void convertSigned12(Vector4f &cf, Vector4s &cs);
void writeDestination(Vector4s &d, const Dst &dst); void writeDestination(Vector4s &d, const Dst &dst);
Vector4s fetchRegisterS(const Src &src); Vector4s fetchRegister(const Src &src);
// Instructions // Instructions
void MOV(Vector4s &dst, Vector4s &src0); void MOV(Vector4s &dst, Vector4s &src0);
......
src/Shader/PixelProgram.cpp
...@@ -136,11 +136,11 @@ namespace sw ...@@ -136,11 +136,11 @@ namespace sw
} }
} }
if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegisterF(src0); if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegister(src0);
if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegisterF(src1); if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegister(src1);
if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegisterF(src2); if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegister(src2);
if(src3.type != Shader::PARAMETER_VOID) s3 = fetchRegisterF(src3); if(src3.type != Shader::PARAMETER_VOID) s3 = fetchRegister(src3);
if(src4.type != Shader::PARAMETER_VOID) s4 = fetchRegisterF(src4); if(src4.type != Shader::PARAMETER_VOID) s4 = fetchRegister(src4);
switch(opcode) switch(opcode)
{ {
...@@ -270,8 +270,8 @@ namespace sw ...@@ -270,8 +270,8 @@ namespace sw
case Shader::OPCODE_M3X3: M3X3(d, s0, src1); break; case Shader::OPCODE_M3X3: M3X3(d, s0, src1); break;
case Shader::OPCODE_M3X2: M3X2(d, s0, src1); break; case Shader::OPCODE_M3X2: M3X2(d, s0, src1); break;
case Shader::OPCODE_TEX: TEXLD(d, s0, src1, project, bias); break; case Shader::OPCODE_TEX: TEXLD(d, s0, src1, project, bias); break;
case Shader::OPCODE_TEXLDD: TEXLDD(d, s0, src1, s2, s3, project, bias); break; case Shader::OPCODE_TEXLDD: TEXLDD(d, s0, src1, s2, s3, project); break;
case Shader::OPCODE_TEXLDL: TEXLDL(d, s0, src1, project, bias); break; case Shader::OPCODE_TEXLDL: TEXLDL(d, s0, src1, project); break;
case Shader::OPCODE_TEXSIZE: TEXSIZE(d, s0.x, src1); break; case Shader::OPCODE_TEXSIZE: TEXSIZE(d, s0.x, src1); break;
case Shader::OPCODE_TEXKILL: TEXKILL(cMask, d, dst.mask); break; case Shader::OPCODE_TEXKILL: TEXKILL(cMask, d, dst.mask); break;
case Shader::OPCODE_TEXOFFSET: TEXOFFSET(d, s0, src1, s2, s3, project, bias); break; case Shader::OPCODE_TEXOFFSET: TEXOFFSET(d, s0, src1, s2, s3, project, bias); break;
...@@ -624,15 +624,17 @@ namespace sw ...@@ -624,15 +624,17 @@ namespace sw
} }
} }
void PixelProgram::sampleTexture(Vector4f &c, const Src &sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, bool bias, bool gradients, bool lodProvided) void PixelProgram::sampleTexture(Vector4f &c, const Src &sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, SamplerMethod method)
{ {
Vector4f tmp;
if(sampler.type == Shader::PARAMETER_SAMPLER && sampler.rel.type == Shader::PARAMETER_VOID) if(sampler.type == Shader::PARAMETER_SAMPLER && sampler.rel.type == Shader::PARAMETER_VOID)
{ {
sampleTexture(c, sampler.index, u, v, w, q, dsx, dsy, project, bias, gradients, lodProvided); sampleTexture(tmp, sampler.index, u, v, w, q, dsx, dsy, project, method);
} }
else else
{ {
Int index = As<Int>(Float(fetchRegisterF(sampler).x.x)); Int index = As<Int>(Float(fetchRegister(sampler).x.x));
for(int i = 0; i < TEXTURE_IMAGE_UNITS; i++) for(int i = 0; i < TEXTURE_IMAGE_UNITS; i++)
{ {
...@@ -640,25 +642,30 @@ namespace sw ...@@ -640,25 +642,30 @@ namespace sw
{ {
If(index == i) If(index == i)
{ {
sampleTexture(c, i, u, v, w, q, dsx, dsy, project, bias, gradients, lodProvided); sampleTexture(tmp, i, u, v, w, q, dsx, dsy, project, method);
// FIXME: When the sampler states are the same, we could use one sampler and just index the texture // FIXME: When the sampler states are the same, we could use one sampler and just index the texture
} }
} }
} }
} }
c.x = tmp[(sampler.swizzle >> 0) & 0x3];
c.y = tmp[(sampler.swizzle >> 2) & 0x3];
c.z = tmp[(sampler.swizzle >> 4) & 0x3];
c.w = tmp[(sampler.swizzle >> 6) & 0x3];
} }
void PixelProgram::sampleTexture(Vector4f &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, bool bias, bool gradients, bool lodProvided) void PixelProgram::sampleTexture(Vector4f &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, SamplerMethod method)
{ {
#if PERF_PROFILE #if PERF_PROFILE
Long texTime = Ticks(); Long texTime = Ticks();
#endif #endif
Pointer<Byte> texture = data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture); Pointer<Byte> texture = data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture);
if(!project) if(!project)
{ {
sampler[stage]->sampleTexture(texture, c, u, v, w, q, dsx, dsy, bias, gradients, lodProvided); sampler[stage]->sampleTexture(texture, c, u, v, w, q, dsx, dsy, method);
} }
else else
{ {
...@@ -668,12 +675,12 @@ namespace sw ...@@ -668,12 +675,12 @@ namespace sw
Float4 v_q = v * rq; Float4 v_q = v * rq;
Float4 w_q = w * rq; Float4 w_q = w * rq;
sampler[stage]->sampleTexture(texture, c, u_q, v_q, w_q, q, dsx, dsy, bias, gradients, lodProvided); sampler[stage]->sampleTexture(texture, c, u_q, v_q, w_q, q, dsx, dsy, method);
} }
#if PERF_PROFILE #if PERF_PROFILE
cycles[PERF_TEX] += Ticks() - texTime; cycles[PERF_TEX] += Ticks() - texTime;
#endif #endif
} }
void PixelProgram::clampColor(Vector4f oC[RENDERTARGETS]) void PixelProgram::clampColor(Vector4f oC[RENDERTARGETS])
...@@ -743,7 +750,7 @@ namespace sw ...@@ -743,7 +750,7 @@ namespace sw
return enable; return enable;
} }
Vector4f PixelProgram::fetchRegisterF(const Src &src, unsigned int offset) Vector4f PixelProgram::fetchRegister(const Src &src, unsigned int offset)
{ {
Vector4f reg; Vector4f reg;
unsigned int i = src.index + offset; unsigned int i = src.index + offset;
...@@ -989,8 +996,8 @@ namespace sw ...@@ -989,8 +996,8 @@ namespace sw
void PixelProgram::M3X2(Vector4f &dst, Vector4f &src0, const Src &src1) void PixelProgram::M3X2(Vector4f &dst, Vector4f &src0, const Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
dst.x = dot3(src0, row0); dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1); dst.y = dot3(src0, row1);
...@@ -998,9 +1005,9 @@ namespace sw ...@@ -998,9 +1005,9 @@ namespace sw
void PixelProgram::M3X3(Vector4f &dst, Vector4f &src0, const Src &src1) void PixelProgram::M3X3(Vector4f &dst, Vector4f &src0, const Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
dst.x = dot3(src0, row0); dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1); dst.y = dot3(src0, row1);
...@@ -1009,10 +1016,10 @@ namespace sw ...@@ -1009,10 +1016,10 @@ namespace sw
void PixelProgram::M3X4(Vector4f &dst, Vector4f &src0, const Src &src1) void PixelProgram::M3X4(Vector4f &dst, Vector4f &src0, const Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
Vector4f row3 = fetchRegisterF(src1, 3); Vector4f row3 = fetchRegister(src1, 3);
dst.x = dot3(src0, row0); dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1); dst.y = dot3(src0, row1);
...@@ -1022,9 +1029,9 @@ namespace sw ...@@ -1022,9 +1029,9 @@ namespace sw
void PixelProgram::M4X3(Vector4f &dst, Vector4f &src0, const Src &src1) void PixelProgram::M4X3(Vector4f &dst, Vector4f &src0, const Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
dst.x = dot4(src0, row0); dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1); dst.y = dot4(src0, row1);
...@@ -1033,10 +1040,10 @@ namespace sw ...@@ -1033,10 +1040,10 @@ namespace sw
void PixelProgram::M4X4(Vector4f &dst, Vector4f &src0, const Src &src1) void PixelProgram::M4X4(Vector4f &dst, Vector4f &src0, const Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
Vector4f row3 = fetchRegisterF(src1, 3); Vector4f row3 = fetchRegister(src1, 3);
dst.x = dot4(src0, row0); dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1); dst.y = dot4(src0, row1);
...@@ -1046,13 +1053,7 @@ namespace sw ...@@ -1046,13 +1053,7 @@ namespace sw
void PixelProgram::TEXLD(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias) void PixelProgram::TEXLD(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias)
{ {
Vector4f tmp; sampleTexture(dst, src1, src0.x, src0.y, src0.z, src0.w, src0, src0, project, bias ? Bias : Implicit);
sampleTexture(tmp, src1, src0.x, src0.y, src0.z, src0.w, src0, src0, project, bias);
dst.x = tmp[(src1.swizzle >> 0) & 0x3];
dst.y = tmp[(src1.swizzle >> 2) & 0x3];
dst.z = tmp[(src1.swizzle >> 4) & 0x3];
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
} }
void PixelProgram::TEXOFFSET(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3, bool project, bool bias) void PixelProgram::TEXOFFSET(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3, bool project, bool bias)
...@@ -1085,26 +1086,14 @@ namespace sw ...@@ -1085,26 +1086,14 @@ namespace sw
UNIMPLEMENTED(); UNIMPLEMENTED();
} }
void PixelProgram::TEXLDD(Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, Vector4f &src3, bool project, bool bias) void PixelProgram::TEXLDD(Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, Vector4f &src3, bool project)
{ {
Vector4f tmp; sampleTexture(dst, src1, src0.x, src0.y, src0.z, src0.w, src2, src3, project, Grad);
sampleTexture(tmp, src1, src0.x, src0.y, src0.z, src0.w, src2, src3, project, bias, true);
dst.x = tmp[(src1.swizzle >> 0) & 0x3];
dst.y = tmp[(src1.swizzle >> 2) & 0x3];
dst.z = tmp[(src1.swizzle >> 4) & 0x3];
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
} }
void PixelProgram::TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias) void PixelProgram::TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, bool project)
{ {
Vector4f tmp; sampleTexture(dst, src1, src0.x, src0.y, src0.z, src0.w, src0, src0, project, Lod);
sampleTexture(tmp, src1, src0.x, src0.y, src0.z, src0.w, src0, src0, project, bias, false, true);
dst.x = tmp[(src1.swizzle >> 0) & 0x3];
dst.y = tmp[(src1.swizzle >> 2) & 0x3];
dst.z = tmp[(src1.swizzle >> 4) & 0x3];
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
} }
void PixelProgram::TEXSIZE(Vector4f &dst, Float4 &lod, const Src &src1) void PixelProgram::TEXSIZE(Vector4f &dst, Float4 &lod, const Src &src1)
...@@ -1458,7 +1447,7 @@ namespace sw ...@@ -1458,7 +1447,7 @@ namespace sw
} }
else else
{ {
Int4 condition = As<Int4>(fetchRegisterF(src).x); Int4 condition = As<Int4>(fetchRegister(src).x);
IF(condition); IF(condition);
} }
} }
...@@ -1627,7 +1616,7 @@ namespace sw ...@@ -1627,7 +1616,7 @@ namespace sw
Nucleus::setInsertBlock(testBlock); Nucleus::setInsertBlock(testBlock);
enableContinue = restoreContinue; enableContinue = restoreContinue;
const Vector4f &src = fetchRegisterF(temporaryRegister); const Vector4f &src = fetchRegister(temporaryRegister);
Int4 condition = As<Int4>(src.x); Int4 condition = As<Int4>(src.x);
condition &= enableStack[enableIndex - 1]; condition &= enableStack[enableIndex - 1];
enableStack[enableIndex] = condition; enableStack[enableIndex] = condition;
......
src/Shader/PixelProgram.hpp
...@@ -13,6 +13,7 @@ ...@@ -13,6 +13,7 @@
#define sw_PixelProgram_hpp #define sw_PixelProgram_hpp
#include "PixelRoutine.hpp" #include "PixelRoutine.hpp"
#include "SamplerCore.hpp"
namespace sw namespace sw
{ {
...@@ -78,15 +79,15 @@ namespace sw ...@@ -78,15 +79,15 @@ namespace sw
Int4 enableContinue; Int4 enableContinue;
Int4 enableLeave; Int4 enableLeave;
void sampleTexture(Vector4f &c, const Src &sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project = false, bool bias = false, bool gradients = false, bool lodProvided = false); void sampleTexture(Vector4f &c, const Src &sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, SamplerMethod method);
void sampleTexture(Vector4f &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project = false, bool bias = false, bool gradients = false, bool lodProvided = false); void sampleTexture(Vector4f &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, SamplerMethod method);
// Raster operations // Raster operations
void clampColor(Vector4f oC[RENDERTARGETS]); void clampColor(Vector4f oC[RENDERTARGETS]);
Int4 enableMask(const Shader::Instruction *instruction); Int4 enableMask(const Shader::Instruction *instruction);
Vector4f fetchRegisterF(const Src &src, unsigned int offset = 0); Vector4f fetchRegister(const Src &src, unsigned int offset = 0);
Vector4f readConstant(const Src &src, unsigned int offset = 0); Vector4f readConstant(const Src &src, unsigned int offset = 0);
RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index); RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index);
RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index, Int& offset); RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index, Int& offset);
...@@ -103,8 +104,8 @@ namespace sw ...@@ -103,8 +104,8 @@ namespace sw
void M4X3(Vector4f &dst, Vector4f &src0, const Src &src1); void M4X3(Vector4f &dst, Vector4f &src0, const Src &src1);
void M4X4(Vector4f &dst, Vector4f &src0, const Src &src1); void M4X4(Vector4f &dst, Vector4f &src0, const Src &src1);
void TEXLD(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias); void TEXLD(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias);
void TEXLDD(Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, Vector4f &src3, bool project, bool bias); void TEXLDD(Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, Vector4f &src3, bool project);
void TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias); void TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, bool project);
void TEXSIZE(Vector4f &dst, Float4 &lod, const Src &src1); void TEXSIZE(Vector4f &dst, Float4 &lod, const Src &src1);
void TEXKILL(Int cMask[4], Vector4f &src, unsigned char mask); void TEXKILL(Int cMask[4], Vector4f &src, unsigned char mask);
void TEXOFFSET(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3, bool project, bool bias); void TEXOFFSET(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3, bool project, bool bias);
......
src/Shader/SamplerCore.cpp
...@@ -51,7 +51,12 @@ namespace sw ...@@ -51,7 +51,12 @@ namespace sw
{ {
} }
void SamplerCore::sampleTexture(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided, bool fixed12) void SamplerCore::sampleTexture(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, SamplerMethod method)
{
sampleTexture(texture, c, u, v, w, q, dsx, dsy, method, true);
}
void SamplerCore::sampleTexture(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, SamplerMethod method, bool fixed12)
{ {
#if PERF_PROFILE #if PERF_PROFILE
AddAtomic(Pointer<Long>(&profiler.texOperations), 4); AddAtomic(Pointer<Long>(&profiler.texOperations), 4);
...@@ -101,16 +106,16 @@ namespace sw ...@@ -101,16 +106,16 @@ namespace sw
{ {
if(state.textureType != TEXTURE_CUBE) if(state.textureType != TEXTURE_CUBE)
{ {
computeLod(texture, lod, anisotropy, uDelta, vDelta, uuuu, vvvv, q.x, dsx, dsy, bias, gradients, lodProvided); computeLod(texture, lod, anisotropy, uDelta, vDelta, uuuu, vvvv, q.x, dsx, dsy, method);
} }
else else
{ {
computeLod(texture, lod, anisotropy, uDelta, vDelta, lodU, lodV, q.x, dsx, dsy, bias, gradients, lodProvided); computeLod(texture, lod, anisotropy, uDelta, vDelta, lodU, lodV, q.x, dsx, dsy, method);
} }
} }
else else
{ {
computeLod3D(texture, lod, uuuu, vvvv, wwww, q.x, dsx, dsy, bias, gradients, lodProvided); computeLod3D(texture, lod, uuuu, vvvv, wwww, q.x, dsx, dsy, method);
} }
if(state.textureType == TEXTURE_CUBE) if(state.textureType == TEXTURE_CUBE)
...@@ -121,13 +126,13 @@ namespace sw ...@@ -121,13 +126,13 @@ namespace sw
if(!hasFloatTexture()) if(!hasFloatTexture())
{ {
sampleFilter(texture, c, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, lodProvided); sampleFilter(texture, c, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, method);
} }
else else
{ {
Vector4f cf; Vector4f cf;
sampleFloatFilter(texture, cf, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, lodProvided); sampleFloatFilter(texture, cf, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, method);
convertFixed12(c, cf); convertFixed12(c, cf);
} }
...@@ -287,7 +292,7 @@ namespace sw ...@@ -287,7 +292,7 @@ namespace sw
} }
} }
void SamplerCore::sampleTexture(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided) void SamplerCore::sampleTexture(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, SamplerMethod method)
{ {
#if PERF_PROFILE #if PERF_PROFILE
AddAtomic(Pointer<Long>(&profiler.texOperations), 4); AddAtomic(Pointer<Long>(&profiler.texOperations), 4);
...@@ -331,16 +336,16 @@ namespace sw ...@@ -331,16 +336,16 @@ namespace sw
{ {
if(state.textureType != TEXTURE_CUBE) if(state.textureType != TEXTURE_CUBE)
{ {
computeLod(texture, lod, anisotropy, uDelta, vDelta, uuuu, vvvv, q.x, dsx, dsy, bias, gradients, lodProvided); computeLod(texture, lod, anisotropy, uDelta, vDelta, uuuu, vvvv, q.x, dsx, dsy, method);
} }
else else
{ {
computeLod(texture, lod, anisotropy, uDelta, vDelta, lodU, lodV, q.x, dsx, dsy, bias, gradients, lodProvided); computeLod(texture, lod, anisotropy, uDelta, vDelta, lodU, lodV, q.x, dsx, dsy, method);
} }
} }
else else
{ {
computeLod3D(texture, lod, uuuu, vvvv, wwww, q.x, dsx, dsy, bias, gradients, lodProvided); computeLod3D(texture, lod, uuuu, vvvv, wwww, q.x, dsx, dsy, method);
} }
if(state.textureType == TEXTURE_CUBE) if(state.textureType == TEXTURE_CUBE)
...@@ -349,13 +354,13 @@ namespace sw ...@@ -349,13 +354,13 @@ namespace sw
vvvv += Float4(0.5f); vvvv += Float4(0.5f);
} }
sampleFloatFilter(texture, c, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, lodProvided); sampleFloatFilter(texture, c, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, method);
} }
else else
{ {
Vector4s cs; Vector4s cs;
sampleTexture(texture, cs, u, v, w, q, dsx, dsy, bias, gradients, lodProvided, false); sampleTexture(texture, cs, u, v, w, q, dsx, dsy, method, false);
for(int component = 0; component < textureComponentCount(); component++) for(int component = 0; component < textureComponentCount(); component++)
{ {
...@@ -593,15 +598,15 @@ namespace sw ...@@ -593,15 +598,15 @@ namespace sw
return uvw; return uvw;
} }
void SamplerCore::sampleFilter(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided) void SamplerCore::sampleFilter(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], SamplerMethod method)
{ {
sampleAniso(texture, c, u, v, w, lod, anisotropy, uDelta, vDelta, face, false, lodProvided); sampleAniso(texture, c, u, v, w, lod, anisotropy, uDelta, vDelta, face, false, method);
if(state.mipmapFilter > MIPMAP_POINT) if(state.mipmapFilter > MIPMAP_POINT)
{ {
Vector4s cc; Vector4s cc;
sampleAniso(texture, cc, u, v, w, lod, anisotropy, uDelta, vDelta, face, true, lodProvided); sampleAniso(texture, cc, u, v, w, lod, anisotropy, uDelta, vDelta, face, true, method);
lod *= Float(1 << 16); lod *= Float(1 << 16);
...@@ -689,9 +694,9 @@ namespace sw ...@@ -689,9 +694,9 @@ namespace sw
} }
} }
void SamplerCore::sampleAniso(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided) void SamplerCore::sampleAniso(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, SamplerMethod method)
{ {
if(state.textureFilter != FILTER_ANISOTROPIC || lodProvided) if(state.textureFilter != FILTER_ANISOTROPIC || method == Lod)
{ {
sampleQuad(texture, c, u, v, w, lod, face, secondLOD); sampleQuad(texture, c, u, v, w, lod, face, secondLOD);
} }
...@@ -1093,15 +1098,15 @@ namespace sw ...@@ -1093,15 +1098,15 @@ namespace sw
} }
} }
void SamplerCore::sampleFloatFilter(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided) void SamplerCore::sampleFloatFilter(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], SamplerMethod method)
{ {
sampleFloatAniso(texture, c, u, v, w, lod, anisotropy, uDelta, vDelta, face, false, lodProvided); sampleFloatAniso(texture, c, u, v, w, lod, anisotropy, uDelta, vDelta, face, false, method);
if(state.mipmapFilter > MIPMAP_POINT) if(state.mipmapFilter > MIPMAP_POINT)
{ {
Vector4f cc; Vector4f cc;
sampleFloatAniso(texture, cc, u, v, w, lod, anisotropy, uDelta, vDelta, face, true, lodProvided); sampleFloatAniso(texture, cc, u, v, w, lod, anisotropy, uDelta, vDelta, face, true, method);
Float4 lod4 = Float4(Frac(lod)); Float4 lod4 = Float4(Frac(lod));
...@@ -1168,9 +1173,9 @@ namespace sw ...@@ -1168,9 +1173,9 @@ namespace sw
} }
} }
void SamplerCore::sampleFloatAniso(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided) void SamplerCore::sampleFloatAniso(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, SamplerMethod method)
{ {
if(state.textureFilter != FILTER_ANISOTROPIC || lodProvided) if(state.textureFilter != FILTER_ANISOTROPIC || method == Lod)
{ {
sampleFloat(texture, c, u, v, w, lod, face, secondLOD); sampleFloat(texture, c, u, v, w, lod, face, secondLOD);
} }
...@@ -1390,13 +1395,13 @@ namespace sw ...@@ -1390,13 +1395,13 @@ namespace sw
} }
} }
void SamplerCore::computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &uuuu, Float4 &vvvv, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided) void SamplerCore::computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &uuuu, Float4 &vvvv, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, SamplerMethod method)
{ {
if(!lodProvided) if(method != Lod)
{ {
Float4 duvdxy; Float4 duvdxy;
if(!gradients) if(method != Grad)
{ {
duvdxy = Float4(uuuu.yz, vvvv.yz) - Float4(uuuu.xx, vvvv.xx); duvdxy = Float4(uuuu.yz, vvvv.yz) - Float4(uuuu.xx, vvvv.xx);
} }
...@@ -1440,7 +1445,7 @@ namespace sw ...@@ -1440,7 +1445,7 @@ namespace sw
lod -= Float(0x3F800000); lod -= Float(0x3F800000);
lod *= As<Float>(Int(0x33800000)); lod *= As<Float>(Int(0x33800000));
if(bias) if(method == Bias)
{ {
lod += lodBias; lod += lodBias;
} }
...@@ -1460,20 +1465,20 @@ namespace sw ...@@ -1460,20 +1465,20 @@ namespace sw
lod = Min(lod, Float(MIPMAP_LEVELS - 2)); // Trilinear accesses lod+1 lod = Min(lod, Float(MIPMAP_LEVELS - 2)); // Trilinear accesses lod+1
} }
void SamplerCore::computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &uuuu, Float4 &vvvv, Float4 &wwww, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided) void SamplerCore::computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &uuuu, Float4 &vvvv, Float4 &wwww, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, SamplerMethod method)
{ {
if(state.mipmapFilter == MIPMAP_NONE) if(state.mipmapFilter == MIPMAP_NONE)
{ {
} }
else // Point and linear filter else // Point and linear filter
{ {
if(!lodProvided) if(method != Lod)
{ {
Float4 dudxy; Float4 dudxy;
Float4 dvdxy; Float4 dvdxy;
Float4 dsdxy; Float4 dsdxy;
if(!gradients) if(method != Grad)
{ {
dudxy = uuuu.ywyw - uuuu; dudxy = uuuu.ywyw - uuuu;
dvdxy = vvvv.ywyw - vvvv; dvdxy = vvvv.ywyw - vvvv;
...@@ -1513,7 +1518,7 @@ namespace sw ...@@ -1513,7 +1518,7 @@ namespace sw
lod -= Float(0x3F800000); lod -= Float(0x3F800000);
lod *= As<Float>(Int(0x33800000)); lod *= As<Float>(Int(0x33800000));
if(bias) if(method == Bias)
{ {
lod += lodBias; lod += lodBias;
} }
......
src/Shader/SamplerCore.hpp
...@@ -17,30 +17,40 @@ ...@@ -17,30 +17,40 @@
namespace sw namespace sw
{ {
enum SamplerMethod
{
Implicit,
Bias,
Lod,
Grad,
};
class SamplerCore class SamplerCore
{ {
public: public:
SamplerCore(Pointer<Byte> &r, const Sampler::State &state); SamplerCore(Pointer<Byte> &r, const Sampler::State &state);
void sampleTexture(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias = false, bool gradients = false, bool lodProvided = false, bool fixed12 = true); void sampleTexture(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, SamplerMethod method = Implicit);
void sampleTexture(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias = false, bool gradients = false, bool lodProvided = false); void sampleTexture(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, SamplerMethod method = Implicit);
private: private:
void sampleTexture(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, SamplerMethod method, bool fixed12);
void border(Short4 &mask, Float4 &coordinates); void border(Short4 &mask, Float4 &coordinates);
void border(Int4 &mask, Float4 &coordinates); void border(Int4 &mask, Float4 &coordinates);
Short4 offsetSample(Short4 &uvw, Pointer<Byte> &mipmap, int halfOffset, bool wrap, int count, Float &lod); Short4 offsetSample(Short4 &uvw, Pointer<Byte> &mipmap, int halfOffset, bool wrap, int count, Float &lod);
void sampleFilter(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided); void sampleFilter(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], SamplerMethod method);
void sampleAniso(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided); void sampleAniso(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, SamplerMethod method);
void sampleQuad(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD); void sampleQuad(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
void sampleQuad2D(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD); void sampleQuad2D(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
void sample3D(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD); void sample3D(Pointer<Byte> &texture, Vector4s &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD);
void sampleFloatFilter(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided); void sampleFloatFilter(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], SamplerMethod method);
void sampleFloatAniso(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided); void sampleFloatAniso(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, SamplerMethod method);
void sampleFloat(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD); void sampleFloat(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
void sampleFloat2D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD); void sampleFloat2D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
void sampleFloat3D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD); void sampleFloat3D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD);
void computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &u, Float4 &v, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided); void computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &u, Float4 &v, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, SamplerMethod method);
void computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &u, Float4 &v, Float4 &w, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided); void computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &u, Float4 &v, Float4 &w, const Float &lodBias, Vector4f &dsx, Vector4f &dsy, SamplerMethod method);
void cubeFace(Int face[4], Float4 &U, Float4 &V, Float4 &lodU, Float4 &lodV, Float4 &x, Float4 &y, Float4 &z); void cubeFace(Int face[4], Float4 &U, Float4 &V, Float4 &lodU, Float4 &lodV, Float4 &x, Float4 &y, Float4 &z);
void computeIndices(Int index[4], Short4 uuuu, Short4 vvvv, Short4 wwww, const Pointer<Byte> &mipmap); void computeIndices(Int index[4], Short4 uuuu, Short4 vvvv, Short4 wwww, const Pointer<Byte> &mipmap);
void sampleTexel(Vector4s &c, Short4 &u, Short4 &v, Short4 &s, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4]); void sampleTexel(Vector4s &c, Short4 &u, Short4 &v, Short4 &s, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4]);
......
src/Shader/VertexProgram.cpp
...@@ -136,11 +136,11 @@ namespace sw ...@@ -136,11 +136,11 @@ namespace sw
Vector4f s3; Vector4f s3;
Vector4f s4; Vector4f s4;
if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegisterF(src0); if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegister(src0);
if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegisterF(src1); if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegister(src1);
if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegisterF(src2); if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegister(src2);
if(src3.type != Shader::PARAMETER_VOID) s3 = fetchRegisterF(src3); if(src3.type != Shader::PARAMETER_VOID) s3 = fetchRegister(src3);
if(src4.type != Shader::PARAMETER_VOID) s4 = fetchRegisterF(src4); if(src4.type != Shader::PARAMETER_VOID) s4 = fetchRegister(src4);
switch(opcode) switch(opcode)
{ {
...@@ -647,7 +647,7 @@ namespace sw ...@@ -647,7 +647,7 @@ namespace sw
} }
} }
Vector4f VertexProgram::fetchRegisterF(const Src &src, unsigned int offset) Vector4f VertexProgram::fetchRegister(const Src &src, unsigned int offset)
{ {
Vector4f reg; Vector4f reg;
unsigned int i = src.index + offset; unsigned int i = src.index + offset;
...@@ -830,7 +830,7 @@ namespace sw ...@@ -830,7 +830,7 @@ namespace sw
if(src.rel.deterministic) if(src.rel.deterministic)
{ {
Int a = relativeAddress(src, src.bufferIndex); Int a = relativeAddress(src, src.bufferIndex);
c.x = c.y = c.z = c.w = *Pointer<Float4>(uniformAddress(src.bufferIndex, i, a)); c.x = c.y = c.z = c.w = *Pointer<Float4>(uniformAddress(src.bufferIndex, i, a));
c.x = c.x.xxxx; c.x = c.x.xxxx;
...@@ -930,8 +930,8 @@ namespace sw ...@@ -930,8 +930,8 @@ namespace sw
void VertexProgram::M3X2(Vector4f &dst, Vector4f &src0, Src &src1) void VertexProgram::M3X2(Vector4f &dst, Vector4f &src0, Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
dst.x = dot3(src0, row0); dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1); dst.y = dot3(src0, row1);
...@@ -939,9 +939,9 @@ namespace sw ...@@ -939,9 +939,9 @@ namespace sw
void VertexProgram::M3X3(Vector4f &dst, Vector4f &src0, Src &src1) void VertexProgram::M3X3(Vector4f &dst, Vector4f &src0, Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
dst.x = dot3(src0, row0); dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1); dst.y = dot3(src0, row1);
...@@ -950,10 +950,10 @@ namespace sw ...@@ -950,10 +950,10 @@ namespace sw
void VertexProgram::M3X4(Vector4f &dst, Vector4f &src0, Src &src1) void VertexProgram::M3X4(Vector4f &dst, Vector4f &src0, Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
Vector4f row3 = fetchRegisterF(src1, 3); Vector4f row3 = fetchRegister(src1, 3);
dst.x = dot3(src0, row0); dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1); dst.y = dot3(src0, row1);
...@@ -963,9 +963,9 @@ namespace sw ...@@ -963,9 +963,9 @@ namespace sw
void VertexProgram::M4X3(Vector4f &dst, Vector4f &src0, Src &src1) void VertexProgram::M4X3(Vector4f &dst, Vector4f &src0, Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
dst.x = dot4(src0, row0); dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1); dst.y = dot4(src0, row1);
...@@ -974,10 +974,10 @@ namespace sw ...@@ -974,10 +974,10 @@ namespace sw
void VertexProgram::M4X4(Vector4f &dst, Vector4f &src0, Src &src1) void VertexProgram::M4X4(Vector4f &dst, Vector4f &src0, Src &src1)
{ {
Vector4f row0 = fetchRegisterF(src1, 0); Vector4f row0 = fetchRegister(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1); Vector4f row1 = fetchRegister(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2); Vector4f row2 = fetchRegister(src1, 2);
Vector4f row3 = fetchRegisterF(src1, 3); Vector4f row3 = fetchRegister(src1, 3);
dst.x = dot4(src0, row0); dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1); dst.y = dot4(src0, row1);
...@@ -1259,7 +1259,7 @@ namespace sw ...@@ -1259,7 +1259,7 @@ namespace sw
} }
else else
{ {
Int4 condition = As<Int4>(fetchRegisterF(src).x); Int4 condition = As<Int4>(fetchRegister(src).x);
IF(condition); IF(condition);
} }
} }
...@@ -1429,7 +1429,7 @@ namespace sw ...@@ -1429,7 +1429,7 @@ namespace sw
Nucleus::setInsertBlock(testBlock); Nucleus::setInsertBlock(testBlock);
enableContinue = restoreContinue; enableContinue = restoreContinue;
const Vector4f &src = fetchRegisterF(temporaryRegister); const Vector4f &src = fetchRegister(temporaryRegister);
Int4 condition = As<Int4>(src.x); Int4 condition = As<Int4>(src.x);
condition &= enableStack[enableIndex - 1]; condition &= enableStack[enableIndex - 1];
enableStack[enableIndex] = condition; enableStack[enableIndex] = condition;
...@@ -1494,25 +1494,13 @@ namespace sw ...@@ -1494,25 +1494,13 @@ namespace sw
void VertexProgram::TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1) void VertexProgram::TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1)
{ {
Vector4f tmp; sampleTexture(dst, src1, src0.x, src0.y, src0.z, src0.w, Lod);
sampleTexture(tmp, src1, src0.x, src0.y, src0.z, src0.w);
dst.x = tmp[(src1.swizzle >> 0) & 0x3];
dst.y = tmp[(src1.swizzle >> 2) & 0x3];
dst.z = tmp[(src1.swizzle >> 4) & 0x3];
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
} }
void VertexProgram::TEX(Vector4f &dst, Vector4f &src0, const Src &src1) void VertexProgram::TEX(Vector4f &dst, Vector4f &src0, const Src &src1)
{ {
Float4 lod = Float4(0.0f); Float4 lod0 = Float4(0.0f);
Vector4f tmp; sampleTexture(dst, src1, src0.x, src0.y, src0.z, lod0, Lod);
sampleTexture(tmp, src1, src0.x, src0.y, src0.z, lod);
dst.x = tmp[(src1.swizzle >> 0) & 0x3];
dst.y = tmp[(src1.swizzle >> 2) & 0x3];
dst.z = tmp[(src1.swizzle >> 4) & 0x3];
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
} }
void VertexProgram::TEXOFFSET(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3) void VertexProgram::TEXOFFSET(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3)
...@@ -1557,29 +1545,36 @@ namespace sw ...@@ -1557,29 +1545,36 @@ namespace sw
} }
} }
void VertexProgram::sampleTexture(Vector4f &c, const Src &s, Float4 &u, Float4 &v, Float4 &w, Float4 &q) void VertexProgram::sampleTexture(Vector4f &c, const Src &s, Float4 &u, Float4 &v, Float4 &w, Float4 &q, SamplerMethod method)
{ {
Vector4f tmp;
if(s.type == Shader::PARAMETER_SAMPLER && s.rel.type == Shader::PARAMETER_VOID) if(s.type == Shader::PARAMETER_SAMPLER && s.rel.type == Shader::PARAMETER_VOID)
{ {
Pointer<Byte> texture = data + OFFSET(DrawData,mipmap[16]) + s.index * sizeof(Texture); Pointer<Byte> texture = data + OFFSET(DrawData,mipmap[TEXTURE_IMAGE_UNITS]) + s.index * sizeof(Texture);
sampler[s.index]->sampleTexture(texture, c, u, v, w, q, a0, a0, false, false, true); sampler[s.index]->sampleTexture(texture, tmp, u, v, w, q, a0, a0, method);
} }
else else
{ {
Int index = As<Int>(Float(fetchRegisterF(s).x.x)); Int index = As<Int>(Float(fetchRegister(s).x.x));
for(int i = 0; i < 16; i++) for(int i = 0; i < VERTEX_TEXTURE_IMAGE_UNITS; i++)
{ {
if(shader->usesSampler(i)) if(shader->usesSampler(i))
{ {
If(index == i) If(index == i)
{ {
Pointer<Byte> texture = data + OFFSET(DrawData,mipmap[16]) + i * sizeof(Texture); Pointer<Byte> texture = data + OFFSET(DrawData,mipmap[TEXTURE_IMAGE_UNITS]) + i * sizeof(Texture);
sampler[i]->sampleTexture(texture, c, u, v, w, q, a0, a0, false, false, true); sampler[i]->sampleTexture(texture, tmp, u, v, w, q, a0, a0, method);
// FIXME: When the sampler states are the same, we could use one sampler and just index the texture // FIXME: When the sampler states are the same, we could use one sampler and just index the texture
} }
} }
} }
} }
c.x = tmp[(s.swizzle >> 0) & 0x3];
c.y = tmp[(s.swizzle >> 2) & 0x3];
c.z = tmp[(s.swizzle >> 4) & 0x3];
c.w = tmp[(s.swizzle >> 6) & 0x3];
} }
} }
src/Shader/VertexProgram.hpp
...@@ -14,6 +14,7 @@ ...@@ -14,6 +14,7 @@
#include "VertexRoutine.hpp" #include "VertexRoutine.hpp"
#include "ShaderCore.hpp" #include "ShaderCore.hpp"
#include "SamplerCore.hpp"
#include "Stream.hpp" #include "Stream.hpp"
#include "Types.hpp" #include "Types.hpp"
...@@ -22,7 +23,6 @@ namespace sw ...@@ -22,7 +23,6 @@ namespace sw
{ {
struct Stream; struct Stream;
class VertexShader; class VertexShader;
class SamplerCore;
class VertexProgram : public VertexRoutine, public ShaderCore class VertexProgram : public VertexRoutine, public ShaderCore
{ {
...@@ -63,7 +63,7 @@ namespace sw ...@@ -63,7 +63,7 @@ namespace sw
void program(); void program();
void passThrough(); void passThrough();
Vector4f fetchRegisterF(const Src &src, unsigned int offset = 0); Vector4f fetchRegister(const Src &src, unsigned int offset = 0);
Vector4f readConstant(const Src &src, unsigned int offset = 0); Vector4f readConstant(const Src &src, unsigned int offset = 0);
RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index); RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index);
RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index, Int& offset); RValue<Pointer<Byte>> uniformAddress(int bufferIndex, unsigned int index, Int& offset);
...@@ -111,7 +111,7 @@ namespace sw ...@@ -111,7 +111,7 @@ namespace sw
void TEXGRAD(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3, Vector4f &src4); void TEXGRAD(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3, Vector4f &src4);
void TEXSIZE(Vector4f &dst, Float4 &lod, const Src&); void TEXSIZE(Vector4f &dst, Float4 &lod, const Src&);
void sampleTexture(Vector4f &c, const Src &s, Float4 &u, Float4 &v, Float4 &w, Float4 &q); void sampleTexture(Vector4f &c, const Src &s, Float4 &u, Float4 &v, Float4 &w, Float4 &q, SamplerMethod method);
SamplerCore *sampler[VERTEX_TEXTURE_IMAGE_UNITS]; SamplerCore *sampler[VERTEX_TEXTURE_IMAGE_UNITS];
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment