Skip to content

S
swiftshader
Commit 4f172c78 by Nicolas Capens
Options

Derive Rasterizer from Function<>.

Bug 22652760 Change-Id: I1a8ff978fdc7ad795e469218a931523d624e7cde Reviewed-on: https://swiftshader-review.googlesource.com/4546Tested-by: 's avatarNicolas Capens <capn@google.com> Reviewed-by: 's avatarAlexis Hétu <sugoi@google.com> Reviewed-by: 's avatarNicolas Capens <capn@google.com>
parent 520acc67
Showing with 1239 additions and 1319 deletions
src/Renderer/PixelProcessor.cpp
......@@ -644,7 +644,7 @@ namespace sw
factor.blendConstant4W[3][1] = blendConstantA;
factor.blendConstant4W[3][2] = blendConstantA;
factor.blendConstant4W[3][3] = blendConstantA;
// FIXME: Compact into generic function // FIXME: Clamp
short invBlendConstantR = iround(65535 * (1 - blendConstant.r));
short invBlendConstantG = iround(65535 * (1 - blendConstant.g));
......@@ -680,12 +680,12 @@ namespace sw
factor.blendConstant4F[1][1] = blendConstant.g;
factor.blendConstant4F[1][2] = blendConstant.g;
factor.blendConstant4F[1][3] = blendConstant.g;
factor.blendConstant4F[2][0] = blendConstant.b;
factor.blendConstant4F[2][1] = blendConstant.b;
factor.blendConstant4F[2][2] = blendConstant.b;
factor.blendConstant4F[2][3] = blendConstant.b;
factor.blendConstant4F[3][0] = blendConstant.a;
factor.blendConstant4F[3][1] = blendConstant.a;
factor.blendConstant4F[3][2] = blendConstant.a;
......@@ -700,12 +700,12 @@ namespace sw
factor.invBlendConstant4F[1][1] = 1 - blendConstant.g;
factor.invBlendConstant4F[1][2] = 1 - blendConstant.g;
factor.invBlendConstant4F[1][3] = 1 - blendConstant.g;
factor.invBlendConstant4F[2][0] = 1 - blendConstant.b;
factor.invBlendConstant4F[2][1] = 1 - blendConstant.b;
factor.invBlendConstant4F[2][2] = 1 - blendConstant.b;
factor.invBlendConstant4F[2][3] = 1 - blendConstant.b;
factor.invBlendConstant4F[3][0] = 1 - blendConstant.a;
factor.invBlendConstant4F[3][1] = 1 - blendConstant.a;
factor.invBlendConstant4F[3][2] = 1 - blendConstant.a;
......@@ -873,7 +873,7 @@ namespace sw
state.depthOverride = context->pixelShader && context->pixelShader->depthOverride();
state.shaderContainsKill = context->pixelShader ? context->pixelShader->containsKill() : false;
if(context->alphaTestActive())
{
state.alphaCompareMode = context->alphaCompareMode;
......@@ -1026,7 +1026,7 @@ namespace sw
if(context->colorActive(color, component))
{
state.color[color].component |= 1 << component;
if(point || flatShading)
{
state.color[color].flat |= 1 << component;
......@@ -1038,7 +1038,7 @@ namespace sw
if(context->fogActive())
{
state.fog.component = true;
if(point)
{
state.fog.flat = true;
......@@ -1096,6 +1096,7 @@ namespace sw
{
const bool integerPipeline = (context->pixelShaderVersion() <= 0x0104);
Rasterizer *generator = nullptr;
if(integerPipeline)
{
generator = new PixelPipeline(state, context->pixelShader);
......@@ -1104,8 +1105,9 @@ namespace sw
{
generator = new PixelProgram(state, context->pixelShader);
}
generator->generate();
routine = generator->getRoutine();
routine = (*generator)(L"PixelRoutine_%0.8X", state.shaderID);
delete generator;
routineCache->add(state, routine);
......
src/Renderer/QuadRasterizer.cpp
......@@ -25,18 +25,6 @@ namespace sw
extern int clusterCount;
QuadRasterizer::Registers::Registers()
{
occlusion = 0;
#if PERF_PROFILE
for(int i = 0; i < PERF_TIMERS; i++)
{
cycles[i] = 0;
}
#endif
}
QuadRasterizer::QuadRasterizer(const PixelProcessor::State &state, const PixelShader *pixelShader) : Rasterizer(state), shader(pixelShader)
{
}
......@@ -47,69 +35,63 @@ namespace sw
void QuadRasterizer::generate()
{
Function<Void(Pointer<Byte>, Int, Int, Pointer<Byte>)> function;
{
#if PERF_PROFILE
Long pixelTime = Ticks();
#endif
Pointer<Byte> primitive(function.Arg<0>());
Int count(function.Arg<1>());
Int cluster(function.Arg<2>());
Pointer<Byte> data(function.Arg<3>());
Registers& r = *createRegisters(shader);
r.constants = *Pointer<Pointer<Byte> >(data + OFFSET(DrawData,constants));
r.cluster = cluster;
r.data = data;
Do
#if PERF_PROFILE
for(int i = 0; i < PERF_TIMERS; i++)
{
r.primitive = primitive;
cycles[i] = 0;
}
Int yMin = *Pointer<Int>(primitive + OFFSET(Primitive,yMin));
Int yMax = *Pointer<Int>(primitive + OFFSET(Primitive,yMax));
Long pixelTime = Ticks();
#endif
Int cluster2 = r.cluster + r.cluster;
yMin += clusterCount * 2 - 2 - cluster2;
yMin &= -clusterCount * 2;
yMin += cluster2;
primitive = Arg<0>();
Int count = Arg<1>();
Int cluster = Arg<2>();
data = Arg<3>();
If(yMin < yMax)
{
rasterize(r, yMin, yMax);
}
constants = *Pointer<Pointer<Byte>>(data + OFFSET(DrawData,constants));
occlusion = 0;
primitive += sizeof(Primitive) * state.multiSample;
count--;
}
Until(count == 0)
Do
{
Int yMin = *Pointer<Int>(primitive + OFFSET(Primitive,yMin));
Int yMax = *Pointer<Int>(primitive + OFFSET(Primitive,yMax));
if(state.occlusionEnabled)
Int cluster2 = cluster + cluster;
yMin += clusterCount * 2 - 2 - cluster2;
yMin &= -clusterCount * 2;
yMin += cluster2;
If(yMin < yMax)
{
UInt clusterOcclusion = *Pointer<UInt>(data + OFFSET(DrawData,occlusion) + 4 * cluster);
clusterOcclusion += r.occlusion;
*Pointer<UInt>(data + OFFSET(DrawData,occlusion) + 4 * cluster) = clusterOcclusion;
rasterize(yMin, yMax);
}
#if PERF_PROFILE
r.cycles[PERF_PIXEL] = Ticks() - pixelTime;
primitive += sizeof(Primitive) * state.multiSample;
count--;
}
Until(count == 0)
for(int i = 0; i < PERF_TIMERS; i++)
{
*Pointer<Long>(data + OFFSET(DrawData,cycles[i]) + 8 * cluster) += r.cycles[i];
}
#endif
if(state.occlusionEnabled)
{
UInt clusterOcclusion = *Pointer<UInt>(data + OFFSET(DrawData,occlusion) + 4 * cluster);
clusterOcclusion += occlusion;
*Pointer<UInt>(data + OFFSET(DrawData,occlusion) + 4 * cluster) = clusterOcclusion;
}
Return();
#if PERF_PROFILE
cycles[PERF_PIXEL] = Ticks() - pixelTime;
delete &r;
}
for(int i = 0; i < PERF_TIMERS; i++)
{
*Pointer<Long>(data + OFFSET(DrawData,cycles[i]) + 8 * cluster) += cycles[i];
}
#endif
routine = function(L"PixelRoutine_%0.8X", state.shaderID);
Return();
}
void QuadRasterizer::rasterize(Registers &r, Int &yMin, Int &yMax)
void QuadRasterizer::rasterize(Int &yMin, Int &yMax)
{
Pointer<Byte> cBuffer[RENDERTARGETS];
Pointer<Byte> zBuffer;
......@@ -119,49 +101,49 @@ namespace sw
{
if(state.colorWriteActive(index))
{
cBuffer[index] = *Pointer<Pointer<Byte> >(r.data + OFFSET(DrawData,colorBuffer[index])) + yMin * *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
cBuffer[index] = *Pointer<Pointer<Byte>>(data + OFFSET(DrawData,colorBuffer[index])) + yMin * *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
}
}
if(state.depthTestActive)
{
zBuffer = *Pointer<Pointer<Byte> >(r.data + OFFSET(DrawData,depthBuffer)) + yMin * *Pointer<Int>(r.data + OFFSET(DrawData,depthPitchB));
zBuffer = *Pointer<Pointer<Byte>>(data + OFFSET(DrawData,depthBuffer)) + yMin * *Pointer<Int>(data + OFFSET(DrawData,depthPitchB));
}
if(state.stencilActive)
{
sBuffer = *Pointer<Pointer<Byte> >(r.data + OFFSET(DrawData,stencilBuffer)) + yMin * *Pointer<Int>(r.data + OFFSET(DrawData,stencilPitchB));
sBuffer = *Pointer<Pointer<Byte>>(data + OFFSET(DrawData,stencilBuffer)) + yMin * *Pointer<Int>(data + OFFSET(DrawData,stencilPitchB));
}
Int y = yMin;
Do
{
Int x0a = Int(*Pointer<Short>(r.primitive + OFFSET(Primitive,outline->left) + (y + 0) * sizeof(Primitive::Span)));
Int x0b = Int(*Pointer<Short>(r.primitive + OFFSET(Primitive,outline->left) + (y + 1) * sizeof(Primitive::Span)));
Int x0a = Int(*Pointer<Short>(primitive + OFFSET(Primitive,outline->left) + (y + 0) * sizeof(Primitive::Span)));
Int x0b = Int(*Pointer<Short>(primitive + OFFSET(Primitive,outline->left) + (y + 1) * sizeof(Primitive::Span)));
Int x0 = Min(x0a, x0b);
for(unsigned int q = 1; q < state.multiSample; q++)
{
x0a = Int(*Pointer<Short>(r.primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->left) + (y + 0) * sizeof(Primitive::Span)));
x0b = Int(*Pointer<Short>(r.primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->left) + (y + 1) * sizeof(Primitive::Span)));
x0a = Int(*Pointer<Short>(primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->left) + (y + 0) * sizeof(Primitive::Span)));
x0b = Int(*Pointer<Short>(primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->left) + (y + 1) * sizeof(Primitive::Span)));
x0 = Min(x0, Min(x0a, x0b));
}
x0 &= 0xFFFFFFFE;
Int x1a = Int(*Pointer<Short>(r.primitive + OFFSET(Primitive,outline->right) + (y + 0) * sizeof(Primitive::Span)));
Int x1b = Int(*Pointer<Short>(r.primitive + OFFSET(Primitive,outline->right) + (y + 1) * sizeof(Primitive::Span)));
Int x1a = Int(*Pointer<Short>(primitive + OFFSET(Primitive,outline->right) + (y + 0) * sizeof(Primitive::Span)));
Int x1b = Int(*Pointer<Short>(primitive + OFFSET(Primitive,outline->right) + (y + 1) * sizeof(Primitive::Span)));
Int x1 = Max(x1a, x1b);
for(unsigned int q = 1; q < state.multiSample; q++)
{
x1a = Int(*Pointer<Short>(r.primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->right) + (y + 0) * sizeof(Primitive::Span)));
x1b = Int(*Pointer<Short>(r.primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->right) + (y + 1) * sizeof(Primitive::Span)));
x1a = Int(*Pointer<Short>(primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->right) + (y + 0) * sizeof(Primitive::Span)));
x1b = Int(*Pointer<Short>(primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->right) + (y + 1) * sizeof(Primitive::Span)));
x1 = Max(x1, Max(x1a, x1b));
}
Float4 yyyy = Float4(Float(y)) + *Pointer<Float4>(r.primitive + OFFSET(Primitive,yQuad), 16);
Float4 yyyy = Float4(Float(y)) + *Pointer<Float4>(primitive + OFFSET(Primitive,yQuad), 16);
if(interpolateZ())
{
......@@ -171,10 +153,10 @@ namespace sw
if(state.multiSample > 1)
{
y -= *Pointer<Float4>(r.constants + OFFSET(Constants,Y) + q * sizeof(float4));
y -= *Pointer<Float4>(constants + OFFSET(Constants,Y) + q * sizeof(float4));
}
r.Dz[q] = *Pointer<Float4>(r.primitive + OFFSET(Primitive,z.C), 16) + y * *Pointer<Float4>(r.primitive + OFFSET(Primitive,z.B), 16);
Dz[q] = *Pointer<Float4>(primitive + OFFSET(Primitive,z.C), 16) + y * *Pointer<Float4>(primitive + OFFSET(Primitive,z.B), 16);
}
}
......@@ -182,7 +164,7 @@ namespace sw
{
if(!state.stencilActive && state.depthTestActive && (state.depthCompareMode == DEPTH_LESSEQUAL || state.depthCompareMode == DEPTH_LESS)) // FIXME: Both modes ok?
{
Float4 xxxx = Float4(Float(x0)) + *Pointer<Float4>(r.primitive + OFFSET(Primitive,xQuad), 16);
Float4 xxxx = Float4(Float(x0)) + *Pointer<Float4>(primitive + OFFSET(Primitive,xQuad), 16);
Pointer<Byte> buffer;
Int pitch;
......@@ -190,7 +172,7 @@ namespace sw
if(!state.quadLayoutDepthBuffer)
{
buffer = zBuffer + 4 * x0;
pitch = *Pointer<Int>(r.data + OFFSET(DrawData,depthPitchB));
pitch = *Pointer<Int>(data + OFFSET(DrawData,depthPitchB));
}
else
{
......@@ -199,7 +181,7 @@ namespace sw
For(Int x = x0, x < x1, x += 2)
{
Float4 z = interpolate(xxxx, r.Dz[0], z, r.primitive + OFFSET(Primitive,z), false, false);
Float4 z = interpolate(xxxx, Dz[0], z, primitive + OFFSET(Primitive,z), false, false);
Float4 zValue;
......@@ -254,7 +236,7 @@ namespace sw
{
if(interpolateW())
{
r.Dw = *Pointer<Float4>(r.primitive + OFFSET(Primitive,w.C), 16) + yyyy * *Pointer<Float4>(r.primitive + OFFSET(Primitive,w.B), 16);
Dw = *Pointer<Float4>(primitive + OFFSET(Primitive,w.C), 16) + yyyy * *Pointer<Float4>(primitive + OFFSET(Primitive,w.B), 16);
}
for(int interpolant = 0; interpolant < 10; interpolant++)
......@@ -263,11 +245,11 @@ namespace sw
{
if(state.interpolant[interpolant].component & (1 << component))
{
r.Dv[interpolant][component] = *Pointer<Float4>(r.primitive + OFFSET(Primitive,V[interpolant][component].C), 16);
Dv[interpolant][component] = *Pointer<Float4>(primitive + OFFSET(Primitive,V[interpolant][component].C), 16);
if(!(state.interpolant[interpolant].flat & (1 << component)))
{
r.Dv[interpolant][component] += yyyy * *Pointer<Float4>(r.primitive + OFFSET(Primitive,V[interpolant][component].B), 16);
Dv[interpolant][component] += yyyy * *Pointer<Float4>(primitive + OFFSET(Primitive,V[interpolant][component].B), 16);
}
}
}
......@@ -275,11 +257,11 @@ namespace sw
if(state.fog.component)
{
r.Df = *Pointer<Float4>(r.primitive + OFFSET(Primitive,f.C), 16);
Df = *Pointer<Float4>(primitive + OFFSET(Primitive,f.C), 16);
if(!state.fog.flat)
{
r.Df += yyyy * *Pointer<Float4>(r.primitive + OFFSET(Primitive,f.B), 16);
Df += yyyy * *Pointer<Float4>(primitive + OFFSET(Primitive,f.B), 16);
}
}
......@@ -288,7 +270,7 @@ namespace sw
for(unsigned int q = 0; q < state.multiSample; q++)
{
xLeft[q] = *Pointer<Short4>(r.primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline) + y * sizeof(Primitive::Span));
xLeft[q] = *Pointer<Short4>(primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline) + y * sizeof(Primitive::Span));
xRight[q] = xLeft[q];
xLeft[q] = Swizzle(xLeft[q], 0xA0) - Short4(1, 2, 1, 2);
......@@ -306,7 +288,7 @@ namespace sw
cMask[q] = SignMask(Pack(mask, mask)) & 0x0000000F;
}
quad(r, cBuffer, zBuffer, sBuffer, cMask, x, y);
quad(cBuffer, zBuffer, sBuffer, cMask, x, y);
}
}
......@@ -314,18 +296,18 @@ namespace sw
{
if(state.colorWriteActive(index))
{
cBuffer[index] += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index])) << (1 + sw::log2(clusterCount)); // FIXME: Precompute
cBuffer[index] += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index])) << (1 + sw::log2(clusterCount)); // FIXME: Precompute
}
}
if(state.depthTestActive)
{
zBuffer += *Pointer<Int>(r.data + OFFSET(DrawData,depthPitchB)) << (1 + sw::log2(clusterCount)); // FIXME: Precompute
zBuffer += *Pointer<Int>(data + OFFSET(DrawData,depthPitchB)) << (1 + sw::log2(clusterCount)); // FIXME: Precompute
}
if(state.stencilActive)
{
sBuffer += *Pointer<Int>(r.data + OFFSET(DrawData,stencilPitchB)) << (1 + sw::log2(clusterCount)); // FIXME: Precompute
sBuffer += *Pointer<Int>(data + OFFSET(DrawData,stencilPitchB)) << (1 + sw::log2(clusterCount)); // FIXME: Precompute
}
y += 2 * clusterCount;
......
src/Renderer/QuadRasterizer.hpp
......@@ -27,31 +27,22 @@ namespace sw
virtual ~QuadRasterizer();
struct Registers
{
Registers();
virtual ~Registers() {};
Pointer<Byte> constants;
Pointer<Byte> primitive;
Pointer<Byte> data;
Pointer<Byte> constants;
Float4 Dz[4];
Float4 Dw;
Float4 Dv[10][4];
Float4 Df;
Pointer<Byte> primitive;
Int cluster;
Pointer<Byte> data;
Float4 Dz[4];
Float4 Dw;
Float4 Dv[10][4];
Float4 Df;
UInt occlusion;
UInt occlusion;
#if PERF_PROFILE
Long cycles[PERF_TIMERS];
Long cycles[PERF_TIMERS];
#endif
};
virtual void quad(Registers &r, Pointer<Byte> cBuffer[4], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y) = 0;
virtual Registers* createRegisters(const PixelShader *shader) = 0;
virtual void quad(Pointer<Byte> cBuffer[4], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y) = 0;
bool interpolateZ() const;
bool interpolateW() const;
......@@ -62,7 +53,7 @@ namespace sw
private:
void generate();
void rasterize(Registers &r, Int &yMin, Int &yMax);
void rasterize(Int &yMin, Int &yMax);
};
}
......
src/Renderer/Rasterizer.cpp
......@@ -20,9 +20,4 @@ namespace sw
Rasterizer::~Rasterizer()
{
}
Routine *Rasterizer::getRoutine()
{
return routine;
}
}
src/Renderer/Rasterizer.hpp
......@@ -19,7 +19,7 @@
namespace sw
{
class Rasterizer
class Rasterizer : public Function<Void(Pointer<Byte>, Int, Int, Pointer<Byte>)>
{
public:
Rasterizer(const PixelProcessor::State &state);
......@@ -27,11 +27,8 @@ namespace sw
virtual ~Rasterizer();
virtual void generate() = 0;
Routine *getRoutine();
protected:
Routine *routine;
const PixelProcessor::State &state;
};
}
......
src/Shader/PixelPipeline.cpp
......@@ -17,24 +17,22 @@ namespace sw
{
extern bool postBlendSRGB;
void PixelPipeline::setBuiltins(PixelRoutine::Registers &rBase, Int &x, Int &y, Float4(&z)[4], Float4 &w)
void PixelPipeline::setBuiltins(Int &x, Int &y, Float4(&z)[4], Float4 &w)
{
Registers& r = *static_cast<Registers*>(&rBase);
if(state.color[0].component & 0x1) r.diffuse.x = convertFixed12(r.v[0].x); else r.diffuse.x = Short4(0x1000);
if(state.color[0].component & 0x2) r.diffuse.y = convertFixed12(r.v[0].y); else r.diffuse.y = Short4(0x1000);
if(state.color[0].component & 0x4) r.diffuse.z = convertFixed12(r.v[0].z); else r.diffuse.z = Short4(0x1000);
if(state.color[0].component & 0x8) r.diffuse.w = convertFixed12(r.v[0].w); else r.diffuse.w = Short4(0x1000);
if(state.color[1].component & 0x1) r.specular.x = convertFixed12(r.v[1].x); else r.specular.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(state.color[1].component & 0x2) r.specular.y = convertFixed12(r.v[1].y); else r.specular.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(state.color[1].component & 0x4) r.specular.z = convertFixed12(r.v[1].z); else r.specular.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(state.color[1].component & 0x8) r.specular.w = convertFixed12(r.v[1].w); else r.specular.w = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(state.color[0].component & 0x1) diffuse.x = convertFixed12(v[0].x); else diffuse.x = Short4(0x1000);
if(state.color[0].component & 0x2) diffuse.y = convertFixed12(v[0].y); else diffuse.y = Short4(0x1000);
if(state.color[0].component & 0x4) diffuse.z = convertFixed12(v[0].z); else diffuse.z = Short4(0x1000);
if(state.color[0].component & 0x8) diffuse.w = convertFixed12(v[0].w); else diffuse.w = Short4(0x1000);
if(state.color[1].component & 0x1) specular.x = convertFixed12(v[1].x); else specular.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(state.color[1].component & 0x2) specular.y = convertFixed12(v[1].y); else specular.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(state.color[1].component & 0x4) specular.z = convertFixed12(v[1].z); else specular.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(state.color[1].component & 0x8) specular.w = convertFixed12(v[1].w); else specular.w = Short4(0x0000, 0x0000, 0x0000, 0x0000);
}
void PixelPipeline::fixedFunction(Registers& r)
void PixelPipeline::fixedFunction()
{
r.current = r.diffuse;
current = diffuse;
Vector4s temp(0x0000, 0x0000, 0x0000, 0x0000);
for(int stage = 0; stage < 8; stage++)
......@@ -48,22 +46,20 @@ namespace sw
if(state.textureStage[stage].usesTexture)
{
sampleTexture(r, texture, stage, stage);
sampleTexture(texture, stage, stage);
}
blendTexture(r, temp, texture, stage);
blendTexture(temp, texture, stage);
}
specularPixel(r.current, r.specular);
specularPixel(current, specular);
}
void PixelPipeline::applyShader(PixelRoutine::Registers &rBase, Int cMask[4])
void PixelPipeline::applyShader(Int cMask[4])
{
Registers& r = *static_cast<Registers*>(&rBase);
if(!shader)
{
fixedFunction(r);
fixedFunction();
return;
}
......@@ -98,14 +94,14 @@ namespace sw
Vector4s s1;
Vector4s s2;
if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegisterS(r, src0);
if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegisterS(r, src1);
if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegisterS(r, src2);
if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegisterS(src0);
if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegisterS(src1);
if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegisterS(src2);
Float4 u = version < 0x0104 ? r.v[2 + dst.index].x : r.v[2 + src0.index].x;
Float4 v = version < 0x0104 ? r.v[2 + dst.index].y : r.v[2 + src0.index].y;
Float4 s = version < 0x0104 ? r.v[2 + dst.index].z : r.v[2 + src0.index].z;
Float4 t = version < 0x0104 ? r.v[2 + dst.index].w : r.v[2 + src0.index].w;
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 z = version < 0x0104 ? v[2 + dst.index].z : v[2 + src0.index].z;
Float4 w = version < 0x0104 ? v[2 + dst.index].w : v[2 + src0.index].w;
switch(opcode)
{
......@@ -129,34 +125,34 @@ namespace sw
case Shader::OPCODE_TEXCOORD:
if(version < 0x0104)
{
TEXCOORD(d, u, v, s, dst.index);
}
TEXCOORD(d, x, y, z, dst.index);
}
else
{
if((src0.swizzle & 0x30) == 0x20) // .xyz
{
TEXCRD(d, u, v, s, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
TEXCRD(d, x, y, z, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
else // .xyw
else // .xwy
{
TEXCRD(d, u, v, t, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
TEXCRD(d, x, y, w, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
}
break;
case Shader::OPCODE_TEXKILL:
if(version < 0x0104)
{
TEXKILL(cMask, u, v, s);
TEXKILL(cMask, x, y, z);
}
else if(version == 0x0104)
{
if(dst.type == Shader::PARAMETER_TEXTURE)
{
TEXKILL(cMask, u, v, s);
TEXKILL(cMask, x, y, z);
}
else
{
TEXKILL(cMask, r.rs[dst.index]);
TEXKILL(cMask, rs[dst.index]);
}
}
else ASSERT(false);
......@@ -164,7 +160,7 @@ namespace sw
case Shader::OPCODE_TEX:
if(version < 0x0104)
{
TEX(r, d, u, v, s, dst.index, false);
TEX(d, x, y, z, dst.index, false);
}
else if(version == 0x0104)
{
......@@ -172,41 +168,41 @@ namespace sw
{
if((src0.swizzle & 0x30) == 0x20) // .xyz
{
TEX(r, d, u, v, s, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
TEX(d, x, y, z, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
else // .xyw
{
TEX(r, d, u, v, t, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
TEX(d, x, y, w, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
}
else
{
TEXLD(r, d, s0, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
TEXLD(d, s0, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
}
else ASSERT(false);
break;
case Shader::OPCODE_TEXBEM: TEXBEM(r, d, s0, u, v, s, dst.index); break;
case Shader::OPCODE_TEXBEML: TEXBEML(r, d, s0, u, v, s, dst.index); break;
case Shader::OPCODE_TEXREG2AR: TEXREG2AR(r, d, s0, dst.index); break;
case Shader::OPCODE_TEXREG2GB: TEXREG2GB(r, d, s0, dst.index); break;
case Shader::OPCODE_TEXM3X2PAD: TEXM3X2PAD(r, u, v, s, s0, 0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X2TEX: TEXM3X2TEX(r, d, u, v, s, dst.index, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X3PAD: TEXM3X3PAD(r, u, v, s, s0, pad++ % 2, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X3TEX: TEXM3X3TEX(r, d, u, v, s, dst.index, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X3SPEC: TEXM3X3SPEC(r, d, u, v, s, dst.index, s0, s1); break;
case Shader::OPCODE_TEXM3X3VSPEC: TEXM3X3VSPEC(r, d, u, v, s, dst.index, s0); break;
case Shader::OPCODE_CND: CND(d, s0, s1, s2); break;
case Shader::OPCODE_TEXREG2RGB: TEXREG2RGB(r, d, s0, dst.index); break;
case Shader::OPCODE_TEXDP3TEX: TEXDP3TEX(r, d, u, v, s, dst.index, s0); break;
case Shader::OPCODE_TEXM3X2DEPTH: TEXM3X2DEPTH(r, d, u, v, s, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXDP3: TEXDP3(r, d, u, v, s, s0); break;
case Shader::OPCODE_TEXM3X3: TEXM3X3(r, d, u, v, s, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXDEPTH: TEXDEPTH(r); break;
case Shader::OPCODE_CMP0: CMP(d, s0, s1, s2); break;
case Shader::OPCODE_BEM: BEM(r, d, s0, s1, dst.index); break;
case Shader::OPCODE_PHASE: break;
case Shader::OPCODE_END: break;
case Shader::OPCODE_TEXBEM: TEXBEM(d, s0, x, y, z, dst.index); break;
case Shader::OPCODE_TEXBEML: TEXBEML(d, s0, x, y, z, dst.index); break;
case Shader::OPCODE_TEXREG2AR: TEXREG2AR(d, s0, dst.index); break;
case Shader::OPCODE_TEXREG2GB: TEXREG2GB(d, s0, dst.index); break;
case Shader::OPCODE_TEXM3X2PAD: TEXM3X2PAD(x, y, z, s0, 0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X2TEX: TEXM3X2TEX(d, x, y, z, dst.index, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X3PAD: TEXM3X3PAD(x, y, z, s0, pad++ % 2, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X3TEX: TEXM3X3TEX(d, x, y, z, dst.index, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXM3X3SPEC: TEXM3X3SPEC(d, x, y, z, dst.index, s0, s1); break;
case Shader::OPCODE_TEXM3X3VSPEC: TEXM3X3VSPEC(d, x, y, z, dst.index, s0); break;
case Shader::OPCODE_CND: CND(d, s0, s1, s2); break;
case Shader::OPCODE_TEXREG2RGB: TEXREG2RGB(d, s0, dst.index); break;
case Shader::OPCODE_TEXDP3TEX: TEXDP3TEX(d, x, y, z, dst.index, s0); break;
case Shader::OPCODE_TEXM3X2DEPTH: TEXM3X2DEPTH(d, x, y, z, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXDP3: TEXDP3(d, x, y, z, s0); break;
case Shader::OPCODE_TEXM3X3: TEXM3X3(d, x, y, z, s0, src0.modifier == Shader::MODIFIER_SIGN); break;
case Shader::OPCODE_TEXDEPTH: TEXDEPTH(); break;
case Shader::OPCODE_CMP0: CMP(d, s0, s1, s2); break;
case Shader::OPCODE_BEM: BEM(d, s0, s1, dst.index); break;
case Shader::OPCODE_PHASE: break;
case Shader::OPCODE_END: break;
default:
ASSERT(false);
}
......@@ -248,25 +244,23 @@ namespace sw
{
const Dst &dst = shader->getInstruction(i - 1)->dst;
writeDestination(r, dPairing, dst);
writeDestination(dPairing, dst);
}
if(!pairing)
{
writeDestination(r, d, dst);
writeDestination(d, dst);
}
}
}
}
Bool PixelPipeline::alphaTest(PixelRoutine::Registers &rBase, Int cMask[4])
Bool PixelPipeline::alphaTest(Int cMask[4])
{
Registers& r = *static_cast<Registers*>(&rBase);
r.current.x = Min(r.current.x, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); r.current.x = Max(r.current.x, Short4(0x0000, 0x0000, 0x0000, 0x0000));
r.current.y = Min(r.current.y, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); r.current.y = Max(r.current.y, Short4(0x0000, 0x0000, 0x0000, 0x0000));
r.current.z = Min(r.current.z, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); r.current.z = Max(r.current.z, Short4(0x0000, 0x0000, 0x0000, 0x0000));
r.current.w = Min(r.current.w, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); r.current.w = Max(r.current.w, Short4(0x0000, 0x0000, 0x0000, 0x0000));
current.x = Min(current.x, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.x = Max(current.x, Short4(0x0000, 0x0000, 0x0000, 0x0000));
current.y = Min(current.y, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.y = Max(current.y, Short4(0x0000, 0x0000, 0x0000, 0x0000));
current.z = Min(current.z, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.z = Max(current.z, Short4(0x0000, 0x0000, 0x0000, 0x0000));
current.w = Min(current.w, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.w = Max(current.w, Short4(0x0000, 0x0000, 0x0000, 0x0000));
if(!state.alphaTestActive())
{
......@@ -277,7 +271,7 @@ namespace sw
if(state.transparencyAntialiasing == TRANSPARENCY_NONE)
{
PixelRoutine::alphaTest(r, aMask, r.current.w);
PixelRoutine::alphaTest(aMask, current.w);
for(unsigned int q = 0; q < state.multiSample; q++)
{
......@@ -286,9 +280,9 @@ namespace sw
}
else if(state.transparencyAntialiasing == TRANSPARENCY_ALPHA_TO_COVERAGE)
{
Float4 alpha = Float4(r.current.w) * Float4(1.0f / 0x1000);
Float4 alpha = Float4(current.w) * Float4(1.0f / 0x1000);
alphaToCoverage(r, cMask, alpha);
alphaToCoverage(cMask, alpha);
}
else ASSERT(false);
......@@ -302,10 +296,8 @@ namespace sw
return pass != 0x0;
}
void PixelPipeline::rasterOperation(PixelRoutine::Registers &rBase, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
void PixelPipeline::rasterOperation(Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
{
Registers& r = *static_cast<Registers*>(&rBase);
if(!state.colorWriteActive(0))
{
return;
......@@ -325,53 +317,53 @@ namespace sw
case FORMAT_A16B16G16R16:
if(!postBlendSRGB && state.writeSRGB)
{
linearToSRGB12_16(r, r.current);
linearToSRGB12_16(current);
}
else
{
r.current.x <<= 4;
r.current.y <<= 4;
r.current.z <<= 4;
r.current.w <<= 4;
current.x <<= 4;
current.y <<= 4;
current.z <<= 4;
current.w <<= 4;
}
if(state.targetFormat[0] == FORMAT_R5G6B5)
{
r.current.x &= Short4(0xF800u);
r.current.y &= Short4(0xFC00u);
r.current.z &= Short4(0xF800u);
current.x &= Short4(0xF800u);
current.y &= Short4(0xFC00u);
current.z &= Short4(0xF800u);
}
fogBlend(r, r.current, fog);
fogBlend(current, fog);
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer[0] + q * *Pointer<Int>(r.data + OFFSET(DrawData, colorSliceB[0]));
Vector4s color = r.current;
Pointer<Byte> buffer = cBuffer[0] + q * *Pointer<Int>(data + OFFSET(DrawData, colorSliceB[0]));
Vector4s color = current;
if(state.multiSampleMask & (1 << q))
{
alphaBlend(r, 0, buffer, color, x);
logicOperation(r, 0, buffer, color, x);
writeColor(r, 0, buffer, x, color, sMask[q], zMask[q], cMask[q]);
alphaBlend(0, buffer, color, x);
logicOperation(0, buffer, color, x);
writeColor(0, buffer, x, color, sMask[q], zMask[q], cMask[q]);
}
}
break;
case FORMAT_R32F:
case FORMAT_G32R32F:
case FORMAT_A32B32G32R32F:
convertSigned12(oC, r.current);
PixelRoutine::fogBlend(r, oC, fog);
convertSigned12(oC, current);
PixelRoutine::fogBlend(oC, fog);
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer[0] + q * *Pointer<Int>(r.data + OFFSET(DrawData, colorSliceB[0]));
Pointer<Byte> buffer = cBuffer[0] + q * *Pointer<Int>(data + OFFSET(DrawData, colorSliceB[0]));
Vector4f color = oC;
if(state.multiSampleMask & (1 << q))
{
alphaBlend(r, 0, buffer, color, x);
writeColor(r, 0, buffer, x, color, sMask[q], zMask[q], cMask[q]);
alphaBlend(0, buffer, color, x);
writeColor(0, buffer, x, color, sMask[q], zMask[q], cMask[q]);
}
}
break;
......@@ -380,7 +372,7 @@ namespace sw
}
}
void PixelPipeline::blendTexture(Registers &r, Vector4s &temp, Vector4s &texture, int stage)
void PixelPipeline::blendTexture(Vector4s &temp, Vector4s &texture, int stage)
{
Vector4s *arg1;
Vector4s *arg2;
......@@ -399,10 +391,10 @@ namespace sw
textureStage.thirdArgument == TextureStage::SOURCE_CONSTANT ||
textureStage.thirdArgumentAlpha == TextureStage::SOURCE_CONSTANT)
{
constant.x = *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].constantColor4[0]));
constant.y = *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].constantColor4[1]));
constant.z = *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].constantColor4[2]));
constant.w = *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].constantColor4[3]));
constant.x = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[0]));
constant.y = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[1]));
constant.z = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[2]));
constant.w = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[3]));
}
if(textureStage.firstArgument == TextureStage::SOURCE_TFACTOR ||
......@@ -412,10 +404,10 @@ namespace sw
textureStage.thirdArgument == TextureStage::SOURCE_TFACTOR ||
textureStage.thirdArgumentAlpha == TextureStage::SOURCE_TFACTOR)
{
tfactor.x = *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[0]));
tfactor.y = *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[1]));
tfactor.z = *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[2]));
tfactor.w = *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[3]));
tfactor.x = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[0]));
tfactor.y = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[1]));
tfactor.z = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[2]));
tfactor.w = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]));
}
// Premodulate
......@@ -423,22 +415,22 @@ namespace sw
{
if(state.textureStage[stage - 1].stageOperation == TextureStage::STAGE_PREMODULATE)
{
r.current.x = MulHigh(r.current.x, texture.x) << 4;
r.current.y = MulHigh(r.current.y, texture.y) << 4;
r.current.z = MulHigh(r.current.z, texture.z) << 4;
current.x = MulHigh(current.x, texture.x) << 4;
current.y = MulHigh(current.y, texture.y) << 4;
current.z = MulHigh(current.z, texture.z) << 4;
}
if(state.textureStage[stage - 1].stageOperationAlpha == TextureStage::STAGE_PREMODULATE)
{
r.current.w = MulHigh(r.current.w, texture.w) << 4;
current.w = MulHigh(current.w, texture.w) << 4;
}
}
if(luminance)
{
texture.x = MulHigh(texture.x, r.L) << 4;
texture.y = MulHigh(texture.y, r.L) << 4;
texture.z = MulHigh(texture.z, r.L) << 4;
texture.x = MulHigh(texture.x, L) << 4;
texture.y = MulHigh(texture.y, L) << 4;
texture.z = MulHigh(texture.z, L) << 4;
luminance = false;
}
......@@ -447,9 +439,9 @@ namespace sw
{
case TextureStage::SOURCE_TEXTURE: arg1 = &texture; break;
case TextureStage::SOURCE_CONSTANT: arg1 = &constant; break;
case TextureStage::SOURCE_CURRENT: arg1 = &r.current; break;
case TextureStage::SOURCE_DIFFUSE: arg1 = &r.diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg1 = &r.specular; break;
case TextureStage::SOURCE_CURRENT: arg1 = &current; break;
case TextureStage::SOURCE_DIFFUSE: arg1 = &diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg1 = &specular; break;
case TextureStage::SOURCE_TEMP: arg1 = &temp; break;
case TextureStage::SOURCE_TFACTOR: arg1 = &tfactor; break;
default:
......@@ -460,9 +452,9 @@ namespace sw
{
case TextureStage::SOURCE_TEXTURE: arg2 = &texture; break;
case TextureStage::SOURCE_CONSTANT: arg2 = &constant; break;
case TextureStage::SOURCE_CURRENT: arg2 = &r.current; break;
case TextureStage::SOURCE_DIFFUSE: arg2 = &r.diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg2 = &r.specular; break;
case TextureStage::SOURCE_CURRENT: arg2 = &current; break;
case TextureStage::SOURCE_DIFFUSE: arg2 = &diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg2 = &specular; break;
case TextureStage::SOURCE_TEMP: arg2 = &temp; break;
case TextureStage::SOURCE_TFACTOR: arg2 = &tfactor; break;
default:
......@@ -473,9 +465,9 @@ namespace sw
{
case TextureStage::SOURCE_TEXTURE: arg3 = &texture; break;
case TextureStage::SOURCE_CONSTANT: arg3 = &constant; break;
case TextureStage::SOURCE_CURRENT: arg3 = &r.current; break;
case TextureStage::SOURCE_DIFFUSE: arg3 = &r.diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg3 = &r.specular; break;
case TextureStage::SOURCE_CURRENT: arg3 = &current; break;
case TextureStage::SOURCE_DIFFUSE: arg3 = &diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg3 = &specular; break;
case TextureStage::SOURCE_TEMP: arg3 = &temp; break;
case TextureStage::SOURCE_TFACTOR: arg3 = &tfactor; break;
default:
......@@ -692,19 +684,19 @@ namespace sw
}
break;
case TextureStage::STAGE_BLENDCURRENTALPHA: // Alpha * (Arg1 - Arg2) + Arg2
res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, r.current.w) << 4; res.x = AddSat(res.x, arg2->x);
res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, r.current.w) << 4; res.y = AddSat(res.y, arg2->y);
res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, r.current.w) << 4; res.z = AddSat(res.z, arg2->z);
res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, current.w) << 4; res.x = AddSat(res.x, arg2->x);
res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, current.w) << 4; res.y = AddSat(res.y, arg2->y);
res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, current.w) << 4; res.z = AddSat(res.z, arg2->z);
break;
case TextureStage::STAGE_BLENDDIFFUSEALPHA: // Alpha * (Arg1 - Arg2) + Arg2
res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, r.diffuse.w) << 4; res.x = AddSat(res.x, arg2->x);
res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, r.diffuse.w) << 4; res.y = AddSat(res.y, arg2->y);
res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, r.diffuse.w) << 4; res.z = AddSat(res.z, arg2->z);
res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, diffuse.w) << 4; res.x = AddSat(res.x, arg2->x);
res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, diffuse.w) << 4; res.y = AddSat(res.y, arg2->y);
res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, diffuse.w) << 4; res.z = AddSat(res.z, arg2->z);
break;
case TextureStage::STAGE_BLENDFACTORALPHA: // Alpha * (Arg1 - Arg2) + Arg2
res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.x = AddSat(res.x, arg2->x);
res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.y = AddSat(res.y, arg2->y);
res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.z = AddSat(res.z, arg2->z);
res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.x = AddSat(res.x, arg2->x);
res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.y = AddSat(res.y, arg2->y);
res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.z = AddSat(res.z, arg2->z);
break;
case TextureStage::STAGE_BLENDTEXTUREALPHA: // Alpha * (Arg1 - Arg2) + Arg2
res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, texture.w) << 4; res.x = AddSat(res.x, arg2->x);
......@@ -751,62 +743,62 @@ namespace sw
break;
case TextureStage::STAGE_BUMPENVMAP:
{
r.du = Float4(texture.x) * Float4(1.0f / 0x0FE0);
r.dv = Float4(texture.y) * Float4(1.0f / 0x0FE0);
du = Float4(texture.x) * Float4(1.0f / 0x0FE0);
dv = Float4(texture.y) * Float4(1.0f / 0x0FE0);
Float4 du2;
Float4 dv2;
du2 = r.du;
dv2 = r.dv;
r.du *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
r.du += dv2;
r.dv *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
r.dv += du2;
du2 = du;
dv2 = dv;
du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
du += dv2;
dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
dv += du2;
perturbate = true;
res.x = r.current.x;
res.y = r.current.y;
res.z = r.current.z;
res.w = r.current.w;
res.x = current.x;
res.y = current.y;
res.z = current.z;
res.w = current.w;
}
break;
case TextureStage::STAGE_BUMPENVMAPLUMINANCE:
{
r.du = Float4(texture.x) * Float4(1.0f / 0x0FE0);
r.dv = Float4(texture.y) * Float4(1.0f / 0x0FE0);
du = Float4(texture.x) * Float4(1.0f / 0x0FE0);
dv = Float4(texture.y) * Float4(1.0f / 0x0FE0);
Float4 du2;
Float4 dv2;
du2 = r.du;
dv2 = r.dv;
du2 = du;
dv2 = dv;
r.du *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
r.du += dv2;
r.dv *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
r.dv += du2;
du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
du += dv2;
dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
dv += du2;
perturbate = true;
r.L = texture.z;
r.L = MulHigh(r.L, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].luminanceScale4)));
r.L = r.L << 4;
r.L = AddSat(r.L, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].luminanceOffset4)));
r.L = Max(r.L, Short4(0x0000, 0x0000, 0x0000, 0x0000));
r.L = Min(r.L, Short4(0x1000));
L = texture.z;
L = MulHigh(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceScale4)));
L = L << 4;
L = AddSat(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceOffset4)));
L = Max(L, Short4(0x0000, 0x0000, 0x0000, 0x0000));
L = Min(L, Short4(0x1000));
luminance = true;
res.x = r.current.x;
res.y = r.current.y;
res.z = r.current.z;
res.w = r.current.w;
res.x = current.x;
res.y = current.y;
res.z = current.z;
res.w = current.w;
}
break;
default:
......@@ -819,9 +811,9 @@ namespace sw
{
case TextureStage::SOURCE_TEXTURE: arg1 = &texture; break;
case TextureStage::SOURCE_CONSTANT: arg1 = &constant; break;
case TextureStage::SOURCE_CURRENT: arg1 = &r.current; break;
case TextureStage::SOURCE_DIFFUSE: arg1 = &r.diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg1 = &r.specular; break;
case TextureStage::SOURCE_CURRENT: arg1 = &current; break;
case TextureStage::SOURCE_DIFFUSE: arg1 = &diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg1 = &specular; break;
case TextureStage::SOURCE_TEMP: arg1 = &temp; break;
case TextureStage::SOURCE_TFACTOR: arg1 = &tfactor; break;
default:
......@@ -832,9 +824,9 @@ namespace sw
{
case TextureStage::SOURCE_TEXTURE: arg2 = &texture; break;
case TextureStage::SOURCE_CONSTANT: arg2 = &constant; break;
case TextureStage::SOURCE_CURRENT: arg2 = &r.current; break;
case TextureStage::SOURCE_DIFFUSE: arg2 = &r.diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg2 = &r.specular; break;
case TextureStage::SOURCE_CURRENT: arg2 = &current; break;
case TextureStage::SOURCE_DIFFUSE: arg2 = &diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg2 = &specular; break;
case TextureStage::SOURCE_TEMP: arg2 = &temp; break;
case TextureStage::SOURCE_TFACTOR: arg2 = &tfactor; break;
default:
......@@ -845,9 +837,9 @@ namespace sw
{
case TextureStage::SOURCE_TEXTURE: arg3 = &texture; break;
case TextureStage::SOURCE_CONSTANT: arg3 = &constant; break;
case TextureStage::SOURCE_CURRENT: arg3 = &r.current; break;
case TextureStage::SOURCE_DIFFUSE: arg3 = &r.diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg3 = &r.specular; break;
case TextureStage::SOURCE_CURRENT: arg3 = &current; break;
case TextureStage::SOURCE_DIFFUSE: arg3 = &diffuse; break;
case TextureStage::SOURCE_SPECULAR: arg3 = &specular; break;
case TextureStage::SOURCE_TEMP: arg3 = &temp; break;
case TextureStage::SOURCE_TFACTOR: arg3 = &tfactor; break;
default:
......@@ -970,13 +962,13 @@ namespace sw
case TextureStage::STAGE_DOT3:
break; // Already computed in color channel
case TextureStage::STAGE_BLENDCURRENTALPHA: // Alpha * (Arg1 - Arg2) + Arg2
res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, r.current.w) << 4; res.w = AddSat(res.w, arg2->w);
res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, current.w) << 4; res.w = AddSat(res.w, arg2->w);
break;
case TextureStage::STAGE_BLENDDIFFUSEALPHA: // Arg1 * (Alpha) + Arg2 * (1 - Alpha)
res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, r.diffuse.w) << 4; res.w = AddSat(res.w, arg2->w);
res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, diffuse.w) << 4; res.w = AddSat(res.w, arg2->w);
break;
case TextureStage::STAGE_BLENDFACTORALPHA:
res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, *Pointer<Short4>(r.data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.w = AddSat(res.w, arg2->w);
res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.w = AddSat(res.w, arg2->w);
break;
case TextureStage::STAGE_BLENDTEXTUREALPHA: // Arg1 * (Alpha) + Arg2 * (1 - Alpha)
res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, texture.w) << 4; res.w = AddSat(res.w, arg2->w);
......@@ -1158,10 +1150,10 @@ namespace sw
switch(textureStage.destinationArgument)
{
case TextureStage::DESTINATION_CURRENT:
r.current.x = res.x;
r.current.y = res.y;
r.current.z = res.z;
r.current.w = res.w;
current.x = res.x;
current.y = res.y;
current.z = res.z;
current.w = res.w;
break;
case TextureStage::DESTINATION_TEMP:
temp.x = res.x;
......@@ -1174,7 +1166,7 @@ namespace sw
}
}
void PixelPipeline::fogBlend(Registers &r, Vector4s &current, Float4 &f)
void PixelPipeline::fogBlend(Vector4s &current, Float4 &f)
{
if(!state.fogActive)
{
......@@ -1183,7 +1175,7 @@ namespace sw
if(state.pixelFogMode != FOG_NONE)
{
pixelFog(r, f);
pixelFog(f);
}
UShort4 fog = convertFixed16(f, true);
......@@ -1194,9 +1186,9 @@ namespace sw
UShort4 invFog = UShort4(0xFFFFu) - fog;
current.x += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(r.data + OFFSET(DrawData, fog.color4[0]))));
current.y += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(r.data + OFFSET(DrawData, fog.color4[1]))));
current.z += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(r.data + OFFSET(DrawData, fog.color4[2]))));
current.x += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(data + OFFSET(DrawData, fog.color4[0]))));
current.y += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(data + OFFSET(DrawData, fog.color4[1]))));
current.z += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(data + OFFSET(DrawData, fog.color4[2]))));
}
void PixelPipeline::specularPixel(Vector4s &current, Vector4s &specular)
......@@ -1211,39 +1203,39 @@ namespace sw
current.z = AddSat(current.z, specular.z);
}
void PixelPipeline::sampleTexture(Registers &r, Vector4s &c, int coordinates, int stage, bool project)
void PixelPipeline::sampleTexture(Vector4s &c, int coordinates, int stage, bool project)
{
Float4 u = r.v[2 + coordinates].x;
Float4 v = r.v[2 + coordinates].y;
Float4 w = r.v[2 + coordinates].z;
Float4 q = r.v[2 + coordinates].w;
Float4 x = v[2 + coordinates].x;
Float4 y = v[2 + coordinates].y;
Float4 z = v[2 + coordinates].z;
Float4 w = v[2 + coordinates].w;
if(perturbate)
{
u += r.du;
v += r.dv;
x += du;
y += dv;
perturbate = false;
}
sampleTexture(r, c, stage, u, v, w, q, project);
sampleTexture(c, stage, x, y, z, w, project);
}
void PixelPipeline::sampleTexture(Registers &r, 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, bool bias)
{
Vector4f dsx;
Vector4f dsy;
sampleTexture(r, c, stage, u, v, w, q, dsx, dsy, project, bias, false);
sampleTexture(c, stage, u, v, w, q, dsx, dsy, project, bias, false);
}
void PixelPipeline::sampleTexture(Registers &r, Vector4s &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, bool bias, bool gradients, bool lodProvided)
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 = r.data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture);
Pointer<Byte> texture = data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture);
if(!project)
{
......@@ -1261,7 +1253,7 @@ namespace sw
}
#if PERF_PROFILE
r.cycles[PERF_TEX] += Ticks() - texTime;
cycles[PERF_TEX] += Ticks() - texTime;
#endif
}
......@@ -1291,41 +1283,41 @@ namespace sw
cf.w = convertSigned12(cs.w);
}
void PixelPipeline::writeDestination(Registers &r, Vector4s &d, const Dst &dst)
void PixelPipeline::writeDestination(Vector4s &d, const Dst &dst)
{
switch(dst.type)
{
case Shader::PARAMETER_TEMP:
if(dst.mask & 0x1) r.rs[dst.index].x = d.x;
if(dst.mask & 0x2) r.rs[dst.index].y = d.y;
if(dst.mask & 0x4) r.rs[dst.index].z = d.z;
if(dst.mask & 0x8) r.rs[dst.index].w = d.w;
if(dst.mask & 0x1) rs[dst.index].x = d.x;
if(dst.mask & 0x2) rs[dst.index].y = d.y;
if(dst.mask & 0x4) rs[dst.index].z = d.z;
if(dst.mask & 0x8) rs[dst.index].w = d.w;
break;
case Shader::PARAMETER_INPUT:
if(dst.mask & 0x1) r.vs[dst.index].x = d.x;
if(dst.mask & 0x2) r.vs[dst.index].y = d.y;
if(dst.mask & 0x4) r.vs[dst.index].z = d.z;
if(dst.mask & 0x8) r.vs[dst.index].w = d.w;
if(dst.mask & 0x1) vs[dst.index].x = d.x;
if(dst.mask & 0x2) vs[dst.index].y = d.y;
if(dst.mask & 0x4) vs[dst.index].z = d.z;
if(dst.mask & 0x8) vs[dst.index].w = d.w;
break;
case Shader::PARAMETER_CONST: ASSERT(false); break;
case Shader::PARAMETER_TEXTURE:
if(dst.mask & 0x1) r.ts[dst.index].x = d.x;
if(dst.mask & 0x2) r.ts[dst.index].y = d.y;
if(dst.mask & 0x4) r.ts[dst.index].z = d.z;
if(dst.mask & 0x8) r.ts[dst.index].w = d.w;
if(dst.mask & 0x1) ts[dst.index].x = d.x;
if(dst.mask & 0x2) ts[dst.index].y = d.y;
if(dst.mask & 0x4) ts[dst.index].z = d.z;
if(dst.mask & 0x8) ts[dst.index].w = d.w;
break;
case Shader::PARAMETER_COLOROUT:
if(dst.mask & 0x1) r.vs[dst.index].x = d.x;
if(dst.mask & 0x2) r.vs[dst.index].y = d.y;
if(dst.mask & 0x4) r.vs[dst.index].z = d.z;
if(dst.mask & 0x8) r.vs[dst.index].w = d.w;
if(dst.mask & 0x1) vs[dst.index].x = d.x;
if(dst.mask & 0x2) vs[dst.index].y = d.y;
if(dst.mask & 0x4) vs[dst.index].z = d.z;
if(dst.mask & 0x8) vs[dst.index].w = d.w;
break;
default:
ASSERT(false);
}
}
Vector4s PixelPipeline::fetchRegisterS(Registers &r, const Src &src)
Vector4s PixelPipeline::fetchRegisterS(const Src &src)
{
Vector4s *reg;
int i = src.index;
......@@ -1334,21 +1326,21 @@ namespace sw
if(src.type == Shader::PARAMETER_CONST)
{
c.x = *Pointer<Short4>(r.data + OFFSET(DrawData, ps.cW[i][0]));
c.y = *Pointer<Short4>(r.data + OFFSET(DrawData, ps.cW[i][1]));
c.z = *Pointer<Short4>(r.data + OFFSET(DrawData, ps.cW[i][2]));
c.w = *Pointer<Short4>(r.data + OFFSET(DrawData, ps.cW[i][3]));
c.x = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][0]));
c.y = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][1]));
c.z = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][2]));
c.w = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][3]));
}
switch(src.type)
{
case Shader::PARAMETER_TEMP: reg = &r.rs[i]; break;
case Shader::PARAMETER_INPUT: reg = &r.vs[i]; break;
case Shader::PARAMETER_TEMP: reg = &rs[i]; break;
case Shader::PARAMETER_INPUT: reg = &vs[i]; break;
case Shader::PARAMETER_CONST: reg = &c; break;
case Shader::PARAMETER_TEXTURE: reg = &r.ts[i]; break;
case Shader::PARAMETER_VOID: return r.rs[0]; // Dummy
case Shader::PARAMETER_FLOAT4LITERAL: return r.rs[0]; // Dummy
default: ASSERT(false); return r.rs[0];
case Shader::PARAMETER_TEXTURE: reg = &ts[i]; break;
case Shader::PARAMETER_VOID: return rs[0]; // Dummy
case Shader::PARAMETER_FLOAT4LITERAL: return rs[0]; // Dummy
default: ASSERT(false); return rs[0];
}
const Short4 &x = (*reg)[(src.swizzle >> 0) & 0x3];
......@@ -1637,11 +1629,11 @@ namespace sw
}
}
void PixelPipeline::TEXDP3(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src)
void PixelPipeline::TEXDP3(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src)
{
TEXM3X3PAD(r, u, v, s, src, 0, false);
TEXM3X3PAD(u, v, s, src, 0, false);
Short4 t0 = RoundShort4(r.u_ * Float4(0x1000));
Short4 t0 = RoundShort4(u_ * Float4(0x1000));
dst.x = t0;
dst.y = t0;
......@@ -1649,14 +1641,14 @@ namespace sw
dst.w = t0;
}
void PixelPipeline::TEXDP3TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0)
void PixelPipeline::TEXDP3TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0)
{
TEXM3X3PAD(r, u, v, s, src0, 0, false);
TEXM3X3PAD(u, v, s, src0, 0, false);
r.v_ = Float4(0.0f);
r.w_ = Float4(0.0f);
v_ = Float4(0.0f);
w_ = Float4(0.0f);
sampleTexture(r, dst, stage, r.u_, r.v_, r.w_, r.w_);
sampleTexture(dst, stage, u_, v_, w_, w_);
}
void PixelPipeline::TEXKILL(Int cMask[4], Float4 &u, Float4 &v, Float4 &s)
......@@ -1682,21 +1674,21 @@ namespace sw
}
}
void PixelPipeline::TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int sampler, bool project)
void PixelPipeline::TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int sampler, bool project)
{
sampleTexture(r, dst, sampler, u, v, s, s, project);
sampleTexture(dst, sampler, u, v, s, s, project);
}
void PixelPipeline::TEXLD(Registers &r, Vector4s &dst, Vector4s &src, int sampler, bool project)
void PixelPipeline::TEXLD(Vector4s &dst, Vector4s &src, int sampler, bool project)
{
Float4 u = Float4(src.x) * Float4(1.0f / 0x0FFE);
Float4 v = Float4(src.y) * Float4(1.0f / 0x0FFE);
Float4 s = Float4(src.z) * Float4(1.0f / 0x0FFE);
sampleTexture(r, dst, sampler, u, v, s, s, project);
sampleTexture(dst, sampler, u, v, s, s, project);
}
void PixelPipeline::TEXBEM(Registers &r, Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage)
void PixelPipeline::TEXBEM(Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage)
{
Float4 du = Float4(src.x) * Float4(1.0f / 0x0FFE);
Float4 dv = Float4(src.y) * Float4(1.0f / 0x0FFE);
......@@ -1704,20 +1696,20 @@ namespace sw
Float4 du2 = du;
Float4 dv2 = dv;
du *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
du += dv2;
dv *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
dv += du2;
Float4 u_ = u + du;
Float4 v_ = v + dv;
sampleTexture(r, dst, stage, u_, v_, s, s);
sampleTexture(dst, stage, u_, v_, s, s);
}
void PixelPipeline::TEXBEML(Registers &r, Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage)
void PixelPipeline::TEXBEML(Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage)
{
Float4 du = Float4(src.x) * Float4(1.0f / 0x0FFE);
Float4 dv = Float4(src.y) * Float4(1.0f / 0x0FFE);
......@@ -1725,24 +1717,24 @@ namespace sw
Float4 du2 = du;
Float4 dv2 = dv;
du *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0]));
dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0]));
du += dv2;
dv *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1]));
du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1]));
dv += du2;
Float4 u_ = u + du;
Float4 v_ = v + dv;
sampleTexture(r, dst, stage, u_, v_, s, s);
sampleTexture(dst, stage, u_, v_, s, s);
Short4 L;
L = src.z;
L = MulHigh(L, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].luminanceScale4)));
L = MulHigh(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceScale4)));
L = L << 4;
L = AddSat(L, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].luminanceOffset4)));
L = AddSat(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceOffset4)));
L = Max(L, Short4(0x0000, 0x0000, 0x0000, 0x0000));
L = Min(L, Short4(0x1000));
......@@ -1751,94 +1743,94 @@ namespace sw
dst.z = MulHigh(dst.z, L); dst.z = dst.z << 4;
}
void PixelPipeline::TEXREG2AR(Registers &r, Vector4s &dst, Vector4s &src0, int stage)
void PixelPipeline::TEXREG2AR(Vector4s &dst, Vector4s &src0, int stage)
{
Float4 u = Float4(src0.w) * Float4(1.0f / 0x0FFE);
Float4 v = Float4(src0.x) * Float4(1.0f / 0x0FFE);
Float4 s = Float4(src0.z) * Float4(1.0f / 0x0FFE);
sampleTexture(r, dst, stage, u, v, s, s);
sampleTexture(dst, stage, u, v, s, s);
}
void PixelPipeline::TEXREG2GB(Registers &r, Vector4s &dst, Vector4s &src0, int stage)
void PixelPipeline::TEXREG2GB(Vector4s &dst, Vector4s &src0, int stage)
{
Float4 u = Float4(src0.y) * Float4(1.0f / 0x0FFE);
Float4 v = Float4(src0.z) * Float4(1.0f / 0x0FFE);
Float4 s = v;
sampleTexture(r, dst, stage, u, v, s, s);
sampleTexture(dst, stage, u, v, s, s);
}
void PixelPipeline::TEXREG2RGB(Registers &r, Vector4s &dst, Vector4s &src0, int stage)
void PixelPipeline::TEXREG2RGB(Vector4s &dst, Vector4s &src0, int stage)
{
Float4 u = Float4(src0.x) * Float4(1.0f / 0x0FFE);
Float4 v = Float4(src0.y) * Float4(1.0f / 0x0FFE);
Float4 s = Float4(src0.z) * Float4(1.0f / 0x0FFE);
sampleTexture(r, dst, stage, u, v, s, s);
sampleTexture(dst, stage, u, v, s, s);
}
void PixelPipeline::TEXM3X2DEPTH(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src, bool signedScaling)
void PixelPipeline::TEXM3X2DEPTH(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src, bool signedScaling)
{
TEXM3X2PAD(r, u, v, s, src, 1, signedScaling);
TEXM3X2PAD(u, v, s, src, 1, signedScaling);
// z / w
r.u_ *= Rcp_pp(r.v_); // FIXME: Set result to 1.0 when division by zero
u_ *= Rcp_pp(v_); // FIXME: Set result to 1.0 when division by zero
r.oDepth = r.u_;
oDepth = u_;
}
void PixelPipeline::TEXM3X2PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling)
void PixelPipeline::TEXM3X2PAD(Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling)
{
TEXM3X3PAD(r, u, v, s, src0, component, signedScaling);
TEXM3X3PAD(u, v, s, src0, component, signedScaling);
}
void PixelPipeline::TEXM3X2TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling)
void PixelPipeline::TEXM3X2TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling)
{
TEXM3X2PAD(r, u, v, s, src0, 1, signedScaling);
TEXM3X2PAD(u, v, s, src0, 1, signedScaling);
r.w_ = Float4(0.0f);
w_ = Float4(0.0f);
sampleTexture(r, dst, stage, r.u_, r.v_, r.w_, r.w_);
sampleTexture(dst, stage, u_, v_, w_, w_);
}
void PixelPipeline::TEXM3X3(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, bool signedScaling)
void PixelPipeline::TEXM3X3(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, bool signedScaling)
{
TEXM3X3PAD(r, u, v, s, src0, 2, signedScaling);
TEXM3X3PAD(u, v, s, src0, 2, signedScaling);
dst.x = RoundShort4(r.u_ * Float4(0x1000));
dst.y = RoundShort4(r.v_ * Float4(0x1000));
dst.z = RoundShort4(r.w_ * Float4(0x1000));
dst.x = RoundShort4(u_ * Float4(0x1000));
dst.y = RoundShort4(v_ * Float4(0x1000));
dst.z = RoundShort4(w_ * Float4(0x1000));
dst.w = Short4(0x1000);
}
void PixelPipeline::TEXM3X3PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling)
void PixelPipeline::TEXM3X3PAD(Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling)
{
if(component == 0 || previousScaling != signedScaling) // FIXME: Other source modifiers?
{
r.U = Float4(src0.x);
r.V = Float4(src0.y);
r.W = Float4(src0.z);
U = Float4(src0.x);
V = Float4(src0.y);
W = Float4(src0.z);
previousScaling = signedScaling;
}
Float4 x = r.U * u + r.V * v + r.W * s;
Float4 x = U * u + V * v + W * s;
x *= Float4(1.0f / 0x1000);
switch(component)
{
case 0: r.u_ = x; break;
case 1: r.v_ = x; break;
case 2: r.w_ = x; break;
case 0: u_ = x; break;
case 1: v_ = x; break;
case 2: w_ = x; break;
default: ASSERT(false);
}
}
void PixelPipeline::TEXM3X3SPEC(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, Vector4s &src1)
void PixelPipeline::TEXM3X3SPEC(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, Vector4s &src1)
{
TEXM3X3PAD(r, u, v, s, src0, 2, false);
TEXM3X3PAD(u, v, s, src0, 2, false);
Float4 E[3]; // Eye vector
......@@ -1852,43 +1844,43 @@ namespace sw
Float4 w__;
// (u'', v'', w'') = 2 * (N . E) * N - E * (N . N)
u__ = r.u_ * E[0];
v__ = r.v_ * E[1];
w__ = r.w_ * E[2];
u__ = u_ * E[0];
v__ = v_ * E[1];
w__ = w_ * E[2];
u__ += v__ + w__;
u__ += u__;
v__ = u__;
w__ = u__;
u__ *= r.u_;
v__ *= r.v_;
w__ *= r.w_;
r.u_ *= r.u_;
r.v_ *= r.v_;
r.w_ *= r.w_;
r.u_ += r.v_ + r.w_;
u__ -= E[0] * r.u_;
v__ -= E[1] * r.u_;
w__ -= E[2] * r.u_;
sampleTexture(r, dst, stage, u__, v__, w__, w__);
u__ *= u_;
v__ *= v_;
w__ *= w_;
u_ *= u_;
v_ *= v_;
w_ *= w_;
u_ += v_ + w_;
u__ -= E[0] * u_;
v__ -= E[1] * u_;
w__ -= E[2] * u_;
sampleTexture(dst, stage, u__, v__, w__, w__);
}
void PixelPipeline::TEXM3X3TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling)
void PixelPipeline::TEXM3X3TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling)
{
TEXM3X3PAD(r, u, v, s, src0, 2, signedScaling);
TEXM3X3PAD(u, v, s, src0, 2, signedScaling);
sampleTexture(r, dst, stage, r.u_, r.v_, r.w_, r.w_);
sampleTexture(dst, stage, u_, v_, w_, w_);
}
void PixelPipeline::TEXM3X3VSPEC(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0)
void PixelPipeline::TEXM3X3VSPEC(Vector4s &dst, Float4 &x, Float4 &y, Float4 &z, int stage, Vector4s &src0)
{
TEXM3X3PAD(r, u, v, s, src0, 2, false);
TEXM3X3PAD(x, y, z, src0, 2, false);
Float4 E[3]; // Eye vector
E[0] = r.v[2 + stage - 2].w;
E[1] = r.v[2 + stage - 1].w;
E[2] = r.v[2 + stage - 0].w;
E[0] = v[2 + stage - 2].w;
E[1] = v[2 + stage - 1].w;
E[2] = v[2 + stage - 0].w;
// Reflection
Float4 u__;
......@@ -1896,36 +1888,36 @@ namespace sw
Float4 w__;
// (u'', v'', w'') = 2 * (N . E) * N - E * (N . N)
u__ = r.u_ * E[0];
v__ = r.v_ * E[1];
w__ = r.w_ * E[2];
u__ = u_ * E[0];
v__ = v_ * E[1];
w__ = w_ * E[2];
u__ += v__ + w__;
u__ += u__;
v__ = u__;
w__ = u__;
u__ *= r.u_;
v__ *= r.v_;
w__ *= r.w_;
r.u_ *= r.u_;
r.v_ *= r.v_;
r.w_ *= r.w_;
r.u_ += r.v_ + r.w_;
u__ -= E[0] * r.u_;
v__ -= E[1] * r.u_;
w__ -= E[2] * r.u_;
sampleTexture(r, dst, stage, u__, v__, w__, w__);
u__ *= u_;
v__ *= v_;
w__ *= w_;
u_ *= u_;
v_ *= v_;
w_ *= w_;
u_ += v_ + w_;
u__ -= E[0] * u_;
v__ -= E[1] * u_;
w__ -= E[2] * u_;
sampleTexture(dst, stage, u__, v__, w__, w__);
}
void PixelPipeline::TEXDEPTH(Registers &r)
void PixelPipeline::TEXDEPTH()
{
r.u_ = Float4(r.rs[5].x);
r.v_ = Float4(r.rs[5].y);
u_ = Float4(rs[5].x);
v_ = Float4(rs[5].y);
// z / w
r.u_ *= Rcp_pp(r.v_); // FIXME: Set result to 1.0 when division by zero
u_ *= Rcp_pp(v_); // FIXME: Set result to 1.0 when division by zero
r.oDepth = r.u_;
oDepth = u_;
}
void PixelPipeline::CND(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2)
......@@ -1944,21 +1936,21 @@ namespace sw
{Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.w); Short4 t1; t1 = src2.w; t1 &= t0; t0 = ~t0 & src1.w; t0 |= t1; dst.w = t0; };
}
void PixelPipeline::BEM(Registers &r, Vector4s &dst, Vector4s &src0, Vector4s &src1, int stage)
void PixelPipeline::BEM(Vector4s &dst, Vector4s &src0, Vector4s &src1, int stage)
{
Short4 t0;
Short4 t1;
// dst.x = src0.x + BUMPENVMAT00(stage) * src1.x + BUMPENVMAT10(stage) * src1.y
t0 = MulHigh(src1.x, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[0][0]))); t0 = t0 << 4; // FIXME: Matrix components range? Overflow hazard.
t1 = MulHigh(src1.y, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[1][0]))); t1 = t1 << 4; // FIXME: Matrix components range? Overflow hazard.
t0 = MulHigh(src1.x, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[0][0]))); t0 = t0 << 4; // FIXME: Matrix components range? Overflow hazard.
t1 = MulHigh(src1.y, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[1][0]))); t1 = t1 << 4; // FIXME: Matrix components range? Overflow hazard.
t0 = AddSat(t0, t1);
t0 = AddSat(t0, src0.x);
dst.x = t0;
// dst.y = src0.y + BUMPENVMAT01(stage) * src1.x + BUMPENVMAT11(stage) * src1.y
t0 = MulHigh(src1.x, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[0][1]))); t0 = t0 << 4; // FIXME: Matrix components range? Overflow hazard.
t1 = MulHigh(src1.y, *Pointer<Short4>(r.data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[1][1]))); t1 = t1 << 4; // FIXME: Matrix components range? Overflow hazard.
t0 = MulHigh(src1.x, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[0][1]))); t0 = t0 << 4; // FIXME: Matrix components range? Overflow hazard.
t1 = MulHigh(src1.y, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[1][1]))); t1 = t1 << 4; // FIXME: Matrix components range? Overflow hazard.
t0 = AddSat(t0, t1);
t0 = AddSat(t0, src0.y);
dst.y = t0;
......
src/Shader/PixelPipeline.hpp
......@@ -20,98 +20,92 @@ namespace sw
{
public:
PixelPipeline(const PixelProcessor::State &state, const PixelShader *shader) :
PixelRoutine(state, shader), perturbate(false), luminance(false), previousScaling(false) {}
PixelRoutine(state, shader), current(rs[0]), diffuse(vs[0]), specular(vs[1]), perturbate(false), luminance(false), previousScaling(false) {}
virtual ~PixelPipeline() {}
protected:
virtual void setBuiltins(PixelRoutine::Registers &r, Int &x, Int &y, Float4(&z)[4], Float4 &w);
virtual void applyShader(PixelRoutine::Registers &r, Int cMask[4]);
virtual Bool alphaTest(PixelRoutine::Registers &r, Int cMask[4]);
virtual void rasterOperation(PixelRoutine::Registers &r, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
virtual QuadRasterizer::Registers* createRegisters(const PixelShader *shader) { return new PixelPipeline::Registers(shader); };
virtual void setBuiltins(Int &x, Int &y, Float4(&z)[4], Float4 &w);
virtual void applyShader(Int cMask[4]);
virtual Bool alphaTest(Int cMask[4]);
virtual void rasterOperation(Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
private:
struct Registers : public PixelRoutine::Registers
{
Registers(const PixelShader *shader) : PixelRoutine::Registers(shader), current(rs[0]), diffuse(vs[0]), specular(vs[1]) {}
Vector4s &current;
Vector4s &diffuse;
Vector4s &specular;
Vector4s rs[6];
Vector4s vs[2];
Vector4s ts[6];
// bem(l) offsets and luminance
Float4 du;
Float4 dv;
Short4 L;
// texm3x3 temporaries
Float4 u_; // FIXME
Float4 v_; // FIXME
Float4 w_; // FIXME
Float4 U; // FIXME
Float4 V; // FIXME
Float4 W; // FIXME
};
Vector4s &current;
Vector4s &diffuse;
Vector4s &specular;
void fixedFunction(Registers& r);
void blendTexture(Registers &r, Vector4s &temp, Vector4s &texture, int stage);
void fogBlend(Registers &r, Vector4s &current, Float4 &fog);
Vector4s rs[6];
Vector4s vs[2];
Vector4s ts[6];
// bem(l) offsets and luminance
Float4 du;
Float4 dv;
Short4 L;
// texm3x3 temporaries
Float4 u_; // FIXME
Float4 v_; // FIXME
Float4 w_; // FIXME
Float4 U; // FIXME
Float4 V; // FIXME
Float4 W; // FIXME
void fixedFunction();
void blendTexture(Vector4s &temp, Vector4s &texture, int stage);
void fogBlend(Vector4s &current, Float4 &fog);
void specularPixel(Vector4s &current, Vector4s &specular);
void sampleTexture(Registers &r, Vector4s &c, int coordinates, int sampler, bool project = false);
void sampleTexture(Registers &r, Vector4s &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project = false, bool bias = false);
void sampleTexture(Registers &r, 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);
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, Vector4f &dsx, Vector4f &dsy, bool project = false, bool bias = false, bool gradients = false, bool lodProvided = false);
Short4 convertFixed12(RValue<Float4> cf);
void convertFixed12(Vector4s &cs, Vector4f &cf);
Float4 convertSigned12(Short4 &cs);
Short4 convertFixed12(RValue<Float4> cf);
void convertFixed12(Vector4s &cs, Vector4f &cf);
Float4 convertSigned12(Short4 &cs);
void convertSigned12(Vector4f &cf, Vector4s &cs);
void writeDestination(Registers &r, Vector4s &d, const Dst &dst);
Vector4s fetchRegisterS(Registers &r, const Src &src);
void writeDestination(Vector4s &d, const Dst &dst);
Vector4s fetchRegisterS(const Src &src);
// Instructions
void MOV(Vector4s &dst, Vector4s &src0);
void ADD(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void SUB(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void MAD(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void MUL(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void DP3(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void DP4(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void LRP(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void TEXCOORD(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate);
void TEXCRD(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate, bool project);
void TEXDP3(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src);
void TEXDP3TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0);
void TEXKILL(Int cMask[4], Float4 &u, Float4 &v, Float4 &s);
void TEXKILL(Int cMask[4], Vector4s &dst);
void TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, bool project);
void TEXLD(Registers &r, Vector4s &dst, Vector4s &src, int stage, bool project);
void TEXBEM(Registers &r, Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage);
void TEXBEML(Registers &r, Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage);
void TEXREG2AR(Registers &r, Vector4s &dst, Vector4s &src0, int stage);
void TEXREG2GB(Registers &r, Vector4s &dst, Vector4s &src0, int stage);
void TEXREG2RGB(Registers &r, Vector4s &dst, Vector4s &src0, int stage);
void TEXM3X2DEPTH(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src, bool signedScaling);
void TEXM3X2PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling);
void TEXM3X2TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling);
void TEXM3X3(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, bool signedScaling);
void TEXM3X3PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling);
void TEXM3X3SPEC(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, Vector4s &src1);
void TEXM3X3TEX(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool singedScaling);
void TEXM3X3VSPEC(Registers &r, Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0);
void TEXDEPTH(Registers &r);
void CND(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void CMP(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void BEM(Registers &r, Vector4s &dst, Vector4s &src0, Vector4s &src1, int stage);
// Instructions
void MOV(Vector4s &dst, Vector4s &src0);
void ADD(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void SUB(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void MAD(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void MUL(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void DP3(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void DP4(Vector4s &dst, Vector4s &src0, Vector4s &src1);
void LRP(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void TEXCOORD(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate);
void TEXCRD(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate, bool project);
void TEXDP3(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src);
void TEXDP3TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0);
void TEXKILL(Int cMask[4], Float4 &u, Float4 &v, Float4 &s);
void TEXKILL(Int cMask[4], Vector4s &dst);
void TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, bool project);
void TEXLD(Vector4s &dst, Vector4s &src, int stage, bool project);
void TEXBEM(Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage);
void TEXBEML(Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage);
void TEXREG2AR(Vector4s &dst, Vector4s &src0, int stage);
void TEXREG2GB(Vector4s &dst, Vector4s &src0, int stage);
void TEXREG2RGB(Vector4s &dst, Vector4s &src0, int stage);
void TEXM3X2DEPTH(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src, bool signedScaling);
void TEXM3X2PAD(Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling);
void TEXM3X2TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling);
void TEXM3X3(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, bool signedScaling);
void TEXM3X3PAD(Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling);
void TEXM3X3SPEC(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, Vector4s &src1);
void TEXM3X3TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool singedScaling);
void TEXM3X3VSPEC(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0);
void TEXDEPTH();
void CND(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void CMP(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2);
void BEM(Vector4s &dst, Vector4s &src0, Vector4s &src1, int stage);
bool perturbate;
bool luminance;
bool previousScaling;
bool perturbate;
bool luminance;
bool previousScaling;
};
}
......
src/Shader/PixelProgram.cpp
......@@ -21,62 +21,58 @@ namespace sw
extern bool halfIntegerCoordinates; // Pixel centers are not at integer coordinates
extern bool fullPixelPositionRegister;
void PixelProgram::setBuiltins(PixelRoutine::Registers &rBase, Int &x, Int &y, Float4(&z)[4], Float4 &w)
void PixelProgram::setBuiltins(Int &x, Int &y, Float4(&z)[4], Float4 &w)
{
Registers& r = *static_cast<Registers*>(&rBase);
if(shader->getVersion() >= 0x0300)
{
if(shader->vPosDeclared)
{
if(!halfIntegerCoordinates)
{
r.vPos.x = Float4(Float(x)) + Float4(0, 1, 0, 1);
r.vPos.y = Float4(Float(y)) + Float4(0, 0, 1, 1);
vPos.x = Float4(Float(x)) + Float4(0, 1, 0, 1);
vPos.y = Float4(Float(y)) + Float4(0, 0, 1, 1);
}
else
{
r.vPos.x = Float4(Float(x)) + Float4(0.5f, 1.5f, 0.5f, 1.5f);
r.vPos.y = Float4(Float(y)) + Float4(0.5f, 0.5f, 1.5f, 1.5f);
vPos.x = Float4(Float(x)) + Float4(0.5f, 1.5f, 0.5f, 1.5f);
vPos.y = Float4(Float(y)) + Float4(0.5f, 0.5f, 1.5f, 1.5f);
}
if(fullPixelPositionRegister)
{
r.vPos.z = z[0]; // FIXME: Centroid?
r.vPos.w = w; // FIXME: Centroid?
vPos.z = z[0]; // FIXME: Centroid?
vPos.w = w; // FIXME: Centroid?
}
}
if(shader->vFaceDeclared)
{
Float4 area = *Pointer<Float>(r.primitive + OFFSET(Primitive, area));
Float4 area = *Pointer<Float>(primitive + OFFSET(Primitive, area));
Float4 face = booleanFaceRegister ? Float4(As<Float4>(CmpNLT(area, Float4(0.0f)))) : area;
r.vFace.x = face;
r.vFace.y = face;
r.vFace.z = face;
r.vFace.w = face;
vFace.x = face;
vFace.y = face;
vFace.z = face;
vFace.w = face;
}
}
}
void PixelProgram::applyShader(PixelRoutine::Registers &rBase, Int cMask[4])
void PixelProgram::applyShader(Int cMask[4])
{
Registers& r = *static_cast<Registers*>(&rBase);
r.enableIndex = 0;
r.stackIndex = 0;
enableIndex = 0;
stackIndex = 0;
if(shader->containsLeaveInstruction())
{
r.enableLeave = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
enableLeave = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
for(int i = 0; i < RENDERTARGETS; i++)
{
if(state.targetFormat[i] != FORMAT_NULL)
{
r.oC[i] = Vector4f(0.0f, 0.0f, 0.0f, 0.0f);
oC[i] = Vector4f(0.0f, 0.0f, 0.0f, 0.0f);
}
}
......@@ -129,22 +125,22 @@ namespace sw
{
if(dst.type == Shader::PARAMETER_TEXTURE)
{
d.x = r.v[2 + dst.index].x;
d.y = r.v[2 + dst.index].y;
d.z = r.v[2 + dst.index].z;
d.w = r.v[2 + dst.index].w;
d.x = v[2 + dst.index].x;
d.y = v[2 + dst.index].y;
d.z = v[2 + dst.index].z;
d.w = v[2 + dst.index].w;
}
else
{
d = r.r[dst.index];
d = r[dst.index];
}
}
if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegisterF(r, src0);
if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegisterF(r, src1);
if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegisterF(r, src2);
if(src3.type != Shader::PARAMETER_VOID) s3 = fetchRegisterF(r, src3);
if(src4.type != Shader::PARAMETER_VOID) s4 = fetchRegisterF(r, src4);
if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegisterF(src0);
if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegisterF(src1);
if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegisterF(src2);
if(src3.type != Shader::PARAMETER_VOID) s3 = fetchRegisterF(src3);
if(src4.type != Shader::PARAMETER_VOID) s4 = fetchRegisterF(src4);
switch(opcode)
{
......@@ -268,46 +264,46 @@ namespace sw
case Shader::OPCODE_ACOSH: acosh(d, s0, pp); break;
case Shader::OPCODE_ASINH: asinh(d, s0, pp); break;
case Shader::OPCODE_ATANH: atanh(d, s0, pp); break;
case Shader::OPCODE_M4X4: M4X4(r, d, s0, src1); break;
case Shader::OPCODE_M4X3: M4X3(r, d, s0, src1); break;
case Shader::OPCODE_M3X4: M3X4(r, d, s0, src1); break;
case Shader::OPCODE_M3X3: M3X3(r, d, s0, src1); break;
case Shader::OPCODE_M3X2: M3X2(r, d, s0, src1); break;
case Shader::OPCODE_TEX: TEXLD(r, d, s0, src1, project, bias); break;
case Shader::OPCODE_TEXLDD: TEXLDD(r, d, s0, src1, s2, s3, project, bias); break;
case Shader::OPCODE_TEXLDL: TEXLDL(r, d, s0, src1, project, bias); break;
case Shader::OPCODE_TEXSIZE: TEXSIZE(r, d, s0.x, src1); break;
case Shader::OPCODE_M4X4: M4X4(d, s0, src1); break;
case Shader::OPCODE_M4X3: M4X3(d, s0, src1); break;
case Shader::OPCODE_M3X4: M3X4(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_TEX: TEXLD(d, s0, src1, project, bias); break;
case Shader::OPCODE_TEXLDD: TEXLDD(d, s0, src1, s2, s3, project, bias); break;
case Shader::OPCODE_TEXLDL: TEXLDL(d, s0, src1, project, bias); break;
case Shader::OPCODE_TEXSIZE: TEXSIZE(d, s0.x, src1); break;
case Shader::OPCODE_TEXKILL: TEXKILL(cMask, d, dst.mask); break;
case Shader::OPCODE_TEXOFFSET: TEXOFFSET(r, d, s0, src1, s2, s3, project, bias); break;
case Shader::OPCODE_TEXLDLOFFSET: TEXLDL(r, d, s0, src1, s2, project, bias); break;
case Shader::OPCODE_TEXELFETCH: TEXELFETCH(r, d, s0, src1, s2); break;
case Shader::OPCODE_TEXELFETCHOFFSET: TEXELFETCH(r, d, s0, src1, s2, s3); break;
case Shader::OPCODE_TEXGRAD: TEXGRAD(r, d, s0, src1, s2, s3); break;
case Shader::OPCODE_TEXGRADOFFSET: TEXGRAD(r, d, s0, src1, s2, s3, s4); break;
case Shader::OPCODE_DISCARD: DISCARD(r, cMask, instruction); break;
case Shader::OPCODE_TEXOFFSET: TEXOFFSET(d, s0, src1, s2, s3, project, bias); break;
case Shader::OPCODE_TEXLDLOFFSET: TEXLDL(d, s0, src1, s2, project, bias); break;
case Shader::OPCODE_TEXELFETCH: TEXELFETCH(d, s0, src1, s2); break;
case Shader::OPCODE_TEXELFETCHOFFSET: TEXELFETCH(d, s0, src1, s2, s3); break;
case Shader::OPCODE_TEXGRAD: TEXGRAD(d, s0, src1, s2, s3); break;
case Shader::OPCODE_TEXGRADOFFSET: TEXGRAD(d, s0, src1, s2, s3, s4); break;
case Shader::OPCODE_DISCARD: DISCARD(cMask, instruction); break;
case Shader::OPCODE_DFDX: DFDX(d, s0); break;
case Shader::OPCODE_DFDY: DFDY(d, s0); break;
case Shader::OPCODE_FWIDTH: FWIDTH(d, s0); break;
case Shader::OPCODE_BREAK: BREAK(r); break;
case Shader::OPCODE_BREAKC: BREAKC(r, s0, s1, control); break;
case Shader::OPCODE_BREAKP: BREAKP(r, src0); break;
case Shader::OPCODE_CONTINUE: CONTINUE(r); break;
case Shader::OPCODE_BREAK: BREAK(); break;
case Shader::OPCODE_BREAKC: BREAKC(s0, s1, control); break;
case Shader::OPCODE_BREAKP: BREAKP(src0); break;
case Shader::OPCODE_CONTINUE: CONTINUE(); break;
case Shader::OPCODE_TEST: TEST(); break;
case Shader::OPCODE_CALL: CALL(r, dst.label, dst.callSite); break;
case Shader::OPCODE_CALLNZ: CALLNZ(r, dst.label, dst.callSite, src0); break;
case Shader::OPCODE_ELSE: ELSE(r); break;
case Shader::OPCODE_ENDIF: ENDIF(r); break;
case Shader::OPCODE_ENDLOOP: ENDLOOP(r); break;
case Shader::OPCODE_ENDREP: ENDREP(r); break;
case Shader::OPCODE_ENDWHILE: ENDWHILE(r); break;
case Shader::OPCODE_IF: IF(r, src0); break;
case Shader::OPCODE_IFC: IFC(r, s0, s1, control); break;
case Shader::OPCODE_CALL: CALL(dst.label, dst.callSite); break;
case Shader::OPCODE_CALLNZ: CALLNZ(dst.label, dst.callSite, src0); break;
case Shader::OPCODE_ELSE: ELSE(); break;
case Shader::OPCODE_ENDIF: ENDIF(); break;
case Shader::OPCODE_ENDLOOP: ENDLOOP(); break;
case Shader::OPCODE_ENDREP: ENDREP(); break;
case Shader::OPCODE_ENDWHILE: ENDWHILE(); break;
case Shader::OPCODE_IF: IF(src0); break;
case Shader::OPCODE_IFC: IFC(s0, s1, control); break;
case Shader::OPCODE_LABEL: LABEL(dst.index); break;
case Shader::OPCODE_LOOP: LOOP(r, src1); break;
case Shader::OPCODE_REP: REP(r, src0); break;
case Shader::OPCODE_WHILE: WHILE(r, src0); break;
case Shader::OPCODE_RET: RET(r); break;
case Shader::OPCODE_LEAVE: LEAVE(r); break;
case Shader::OPCODE_LOOP: LOOP(src1); break;
case Shader::OPCODE_REP: REP(src0); break;
case Shader::OPCODE_WHILE: WHILE(src0); break;
case Shader::OPCODE_RET: RET(); break;
case Shader::OPCODE_LEAVE: LEAVE(); break;
case Shader::OPCODE_CMP: cmp(d, s0, s1, control); break;
case Shader::OPCODE_ALL: all(d.x, s0); break;
case Shader::OPCODE_ANY: any(d.x, s0); break;
......@@ -361,53 +357,53 @@ namespace sw
case Shader::PARAMETER_TEMP:
if(dst.rel.type == Shader::PARAMETER_VOID)
{
if(dst.x) pDst.x = r.r[dst.index].x;
if(dst.y) pDst.y = r.r[dst.index].y;
if(dst.z) pDst.z = r.r[dst.index].z;
if(dst.w) pDst.w = r.r[dst.index].w;
if(dst.x) pDst.x = r[dst.index].x;
if(dst.y) pDst.y = r[dst.index].y;
if(dst.z) pDst.z = r[dst.index].z;
if(dst.w) pDst.w = r[dst.index].w;
}
else
{
Int a = relativeAddress(r, dst);
Int a = relativeAddress(dst);
if(dst.x) pDst.x = r.r[dst.index + a].x;
if(dst.y) pDst.y = r.r[dst.index + a].y;
if(dst.z) pDst.z = r.r[dst.index + a].z;
if(dst.w) pDst.w = r.r[dst.index + a].w;
if(dst.x) pDst.x = r[dst.index + a].x;
if(dst.y) pDst.y = r[dst.index + a].y;
if(dst.z) pDst.z = r[dst.index + a].z;
if(dst.w) pDst.w = r[dst.index + a].w;
}
break;
case Shader::PARAMETER_COLOROUT:
if(dst.rel.type == Shader::PARAMETER_VOID)
{
if(dst.x) pDst.x = r.oC[dst.index].x;
if(dst.y) pDst.y = r.oC[dst.index].y;
if(dst.z) pDst.z = r.oC[dst.index].z;
if(dst.w) pDst.w = r.oC[dst.index].w;
if(dst.x) pDst.x = oC[dst.index].x;
if(dst.y) pDst.y = oC[dst.index].y;
if(dst.z) pDst.z = oC[dst.index].z;
if(dst.w) pDst.w = oC[dst.index].w;
}
else
{
Int a = relativeAddress(r, dst) + dst.index;
Int a = relativeAddress(dst) + dst.index;
if(dst.x) pDst.x = r.oC[a].x;
if(dst.y) pDst.y = r.oC[a].y;
if(dst.z) pDst.z = r.oC[a].z;
if(dst.w) pDst.w = r.oC[a].w;
if(dst.x) pDst.x = oC[a].x;
if(dst.y) pDst.y = oC[a].y;
if(dst.z) pDst.z = oC[a].z;
if(dst.w) pDst.w = oC[a].w;
}
break;
case Shader::PARAMETER_PREDICATE:
if(dst.x) pDst.x = r.p0.x;
if(dst.y) pDst.y = r.p0.y;
if(dst.z) pDst.z = r.p0.z;
if(dst.w) pDst.w = r.p0.w;
if(dst.x) pDst.x = p0.x;
if(dst.y) pDst.y = p0.y;
if(dst.z) pDst.z = p0.z;
if(dst.w) pDst.w = p0.w;
break;
case Shader::PARAMETER_DEPTHOUT:
pDst.x = r.oDepth;
pDst.x = oDepth;
break;
default:
ASSERT(false);
}
Int4 enable = enableMask(r, instruction);
Int4 enable = enableMask(instruction);
Int4 xEnable = enable;
Int4 yEnable = enable;
......@@ -418,10 +414,10 @@ namespace sw
{
unsigned char pSwizzle = instruction->predicateSwizzle;
Float4 xPredicate = r.p0[(pSwizzle >> 0) & 0x03];
Float4 yPredicate = r.p0[(pSwizzle >> 2) & 0x03];
Float4 zPredicate = r.p0[(pSwizzle >> 4) & 0x03];
Float4 wPredicate = r.p0[(pSwizzle >> 6) & 0x03];
Float4 xPredicate = p0[(pSwizzle >> 0) & 0x03];
Float4 yPredicate = p0[(pSwizzle >> 2) & 0x03];
Float4 zPredicate = p0[(pSwizzle >> 4) & 0x03];
Float4 wPredicate = p0[(pSwizzle >> 6) & 0x03];
if(!instruction->predicateNot)
{
......@@ -455,47 +451,47 @@ namespace sw
case Shader::PARAMETER_TEMP:
if(dst.rel.type == Shader::PARAMETER_VOID)
{
if(dst.x) r.r[dst.index].x = d.x;
if(dst.y) r.r[dst.index].y = d.y;
if(dst.z) r.r[dst.index].z = d.z;
if(dst.w) r.r[dst.index].w = d.w;
if(dst.x) r[dst.index].x = d.x;
if(dst.y) r[dst.index].y = d.y;
if(dst.z) r[dst.index].z = d.z;
if(dst.w) r[dst.index].w = d.w;
}
else
{
Int a = relativeAddress(r, dst);
Int a = relativeAddress(dst);
if(dst.x) r.r[dst.index + a].x = d.x;
if(dst.y) r.r[dst.index + a].y = d.y;
if(dst.z) r.r[dst.index + a].z = d.z;
if(dst.w) r.r[dst.index + a].w = d.w;
if(dst.x) r[dst.index + a].x = d.x;
if(dst.y) r[dst.index + a].y = d.y;
if(dst.z) r[dst.index + a].z = d.z;
if(dst.w) r[dst.index + a].w = d.w;
}
break;
case Shader::PARAMETER_COLOROUT:
if(dst.rel.type == Shader::PARAMETER_VOID)
{
if(dst.x) { r.oC[dst.index].x = d.x; }
if(dst.y) { r.oC[dst.index].y = d.y; }
if(dst.z) { r.oC[dst.index].z = d.z; }
if(dst.w) { r.oC[dst.index].w = d.w; }
if(dst.x) { oC[dst.index].x = d.x; }
if(dst.y) { oC[dst.index].y = d.y; }
if(dst.z) { oC[dst.index].z = d.z; }
if(dst.w) { oC[dst.index].w = d.w; }
}
else
{
Int a = relativeAddress(r, dst) + dst.index;
Int a = relativeAddress(dst) + dst.index;
if(dst.x) { r.oC[a].x = d.x; }
if(dst.y) { r.oC[a].y = d.y; }
if(dst.z) { r.oC[a].z = d.z; }
if(dst.w) { r.oC[a].w = d.w; }
if(dst.x) { oC[a].x = d.x; }
if(dst.y) { oC[a].y = d.y; }
if(dst.z) { oC[a].z = d.z; }
if(dst.w) { oC[a].w = d.w; }
}
break;
case Shader::PARAMETER_PREDICATE:
if(dst.x) r.p0.x = d.x;
if(dst.y) r.p0.y = d.y;
if(dst.z) r.p0.z = d.z;
if(dst.w) r.p0.w = d.w;
if(dst.x) p0.x = d.x;
if(dst.y) p0.y = d.y;
if(dst.z) p0.z = d.z;
if(dst.w) p0.w = d.w;
break;
case Shader::PARAMETER_DEPTHOUT:
r.oDepth = d.x;
oDepth = d.x;
break;
default:
ASSERT(false);
......@@ -510,15 +506,13 @@ namespace sw
for(int i = 0; i < RENDERTARGETS; i++)
{
r.c[i] = r.oC[i];
c[i] = oC[i];
}
}
Bool PixelProgram::alphaTest(PixelRoutine::Registers &rBase, Int cMask[4])
Bool PixelProgram::alphaTest(Int cMask[4])
{
Registers& r = *static_cast<Registers*>(&rBase);
clampColor(r.c);
clampColor(c);
if(!state.alphaTestActive())
{
......@@ -529,9 +523,9 @@ namespace sw
if(state.transparencyAntialiasing == TRANSPARENCY_NONE)
{
Short4 alpha = RoundShort4(r.c[0].w * Float4(0x1000));
Short4 alpha = RoundShort4(c[0].w * Float4(0x1000));
PixelRoutine::alphaTest(r, aMask, alpha);
PixelRoutine::alphaTest(aMask, alpha);
for(unsigned int q = 0; q < state.multiSample; q++)
{
......@@ -540,7 +534,7 @@ namespace sw
}
else if(state.transparencyAntialiasing == TRANSPARENCY_ALPHA_TO_COVERAGE)
{
alphaToCoverage(r, cMask, r.c[0].w);
alphaToCoverage(cMask, c[0].w);
}
else ASSERT(false);
......@@ -554,10 +548,8 @@ namespace sw
return pass != 0x0;
}
void PixelProgram::rasterOperation(PixelRoutine::Registers &rBase, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
void PixelProgram::rasterOperation(Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
{
Registers& r = *static_cast<Registers*>(&rBase);
for(int index = 0; index < RENDERTARGETS; index++)
{
if(!state.colorWriteActive(index))
......@@ -567,14 +559,14 @@ namespace sw
if(!postBlendSRGB && state.writeSRGB)
{
r.c[index].x = linearToSRGB(r.c[index].x);
r.c[index].y = linearToSRGB(r.c[index].y);
r.c[index].z = linearToSRGB(r.c[index].z);
c[index].x = linearToSRGB(c[index].x);
c[index].y = linearToSRGB(c[index].y);
c[index].z = linearToSRGB(c[index].z);
}
if(index == 0)
{
fogBlend(r, r.c[index], fog);
fogBlend(c[index], fog);
}
switch(state.targetFormat[index])
......@@ -589,19 +581,19 @@ namespace sw
case FORMAT_A16B16G16R16:
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer[index] + q * *Pointer<Int>(r.data + OFFSET(DrawData, colorSliceB[index]));
Pointer<Byte> buffer = cBuffer[index] + q * *Pointer<Int>(data + OFFSET(DrawData, colorSliceB[index]));
Vector4s color;
color.x = convertFixed16(r.c[index].x, false);
color.y = convertFixed16(r.c[index].y, false);
color.z = convertFixed16(r.c[index].z, false);
color.w = convertFixed16(r.c[index].w, false);
color.x = convertFixed16(c[index].x, false);
color.y = convertFixed16(c[index].y, false);
color.z = convertFixed16(c[index].z, false);
color.w = convertFixed16(c[index].w, false);
if(state.multiSampleMask & (1 << q))
{
alphaBlend(r, index, buffer, color, x);
logicOperation(r, index, buffer, color, x);
writeColor(r, index, buffer, x, color, sMask[q], zMask[q], cMask[q]);
alphaBlend(index, buffer, color, x);
logicOperation(index, buffer, color, x);
writeColor(index, buffer, x, color, sMask[q], zMask[q], cMask[q]);
}
}
break;
......@@ -610,13 +602,13 @@ namespace sw
case FORMAT_A32B32G32R32F:
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer[index] + q * *Pointer<Int>(r.data + OFFSET(DrawData, colorSliceB[index]));
Vector4f color = r.c[index];
Pointer<Byte> buffer = cBuffer[index] + q * *Pointer<Int>(data + OFFSET(DrawData, colorSliceB[index]));
Vector4f color = c[index];
if(state.multiSampleMask & (1 << q))
{
alphaBlend(r, index, buffer, color, x);
writeColor(r, index, buffer, x, color, sMask[q], zMask[q], cMask[q]);
alphaBlend(index, buffer, color, x);
writeColor(index, buffer, x, color, sMask[q], zMask[q], cMask[q]);
}
}
break;
......@@ -626,15 +618,15 @@ namespace sw
}
}
void PixelProgram::sampleTexture(Registers &r, 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, bool bias, bool gradients, bool lodProvided)
{
if(sampler.type == Shader::PARAMETER_SAMPLER && sampler.rel.type == Shader::PARAMETER_VOID)
{
sampleTexture(r, c, sampler.index, u, v, w, q, dsx, dsy, project, bias, gradients, lodProvided);
sampleTexture(c, sampler.index, u, v, w, q, dsx, dsy, project, bias, gradients, lodProvided);
}
else
{
Int index = As<Int>(Float(fetchRegisterF(r, sampler).x.x));
Int index = As<Int>(Float(fetchRegisterF(sampler).x.x));
for(int i = 0; i < TEXTURE_IMAGE_UNITS; i++)
{
......@@ -642,7 +634,7 @@ namespace sw
{
If(index == i)
{
sampleTexture(r, c, i, u, v, w, q, dsx, dsy, project, bias, gradients, lodProvided);
sampleTexture(c, i, u, v, w, q, dsx, dsy, project, bias, gradients, lodProvided);
// FIXME: When the sampler states are the same, we could use one sampler and just index the texture
}
}
......@@ -650,13 +642,13 @@ namespace sw
}
}
void PixelProgram::sampleTexture(Registers &r, 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, bool bias, bool gradients, bool lodProvided)
{
#if PERF_PROFILE
Long texTime = Ticks();
#endif
Pointer<Byte> texture = r.data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture);
Pointer<Byte> texture = data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture);
if(!project)
{
......@@ -674,7 +666,7 @@ namespace sw
}
#if PERF_PROFILE
r.cycles[PERF_TEX] += Ticks() - texTime;
cycles[PERF_TEX] += Ticks() - texTime;
#endif
}
......@@ -714,32 +706,32 @@ namespace sw
}
}
Int4 PixelProgram::enableMask(Registers &r, const Shader::Instruction *instruction)
Int4 PixelProgram::enableMask(const Shader::Instruction *instruction)
{
Int4 enable = instruction->analysisBranch ? Int4(r.enableStack[r.enableIndex]) : Int4(0xFFFFFFFF);
Int4 enable = instruction->analysisBranch ? Int4(enableStack[enableIndex]) : Int4(0xFFFFFFFF);
if(!whileTest)
{
if(shader->containsBreakInstruction() && instruction->analysisBreak)
{
enable &= r.enableBreak;
enable &= enableBreak;
}
if(shader->containsContinueInstruction() && instruction->analysisContinue)
{
enable &= r.enableContinue;
enable &= enableContinue;
}
if(shader->containsLeaveInstruction() && instruction->analysisLeave)
{
enable &= r.enableLeave;
enable &= enableLeave;
}
}
return enable;
}
Vector4f PixelProgram::fetchRegisterF(Registers &r, const Src &src, unsigned int offset)
Vector4f PixelProgram::fetchRegisterF(const Src &src, unsigned int offset)
{
Vector4f reg;
unsigned int i = src.index + offset;
......@@ -749,44 +741,44 @@ namespace sw
case Shader::PARAMETER_TEMP:
if(src.rel.type == Shader::PARAMETER_VOID)
{
reg = r.r[i];
reg = r[i];
}
else
{
Int a = relativeAddress(r, src);
Int a = relativeAddress(src);
reg = r.r[i + a];
reg = r[i + a];
}
break;
case Shader::PARAMETER_INPUT:
{
if(src.rel.type == Shader::PARAMETER_VOID) // Not relative
{
reg = r.v[i];
reg = v[i];
}
else if(src.rel.type == Shader::PARAMETER_LOOP)
{
Int aL = r.aL[r.loopDepth];
Int aL = this->aL[loopDepth];
reg = r.v[i + aL];
reg = v[i + aL];
}
else
{
Int a = relativeAddress(r, src);
Int a = relativeAddress(src);
reg = r.v[i + a];
reg = v[i + a];
}
}
break;
case Shader::PARAMETER_CONST:
reg = readConstant(r, src, offset);
reg = readConstant(src, offset);
break;
case Shader::PARAMETER_TEXTURE:
reg = r.v[2 + i];
reg = v[2 + i];
break;
case Shader::PARAMETER_MISCTYPE:
if(src.index == 0) reg = r.vPos;
if(src.index == 1) reg = r.vFace;
if(src.index == 0) reg = vPos;
if(src.index == 1) reg = vFace;
break;
case Shader::PARAMETER_SAMPLER:
if(src.rel.type == Shader::PARAMETER_VOID)
......@@ -795,7 +787,7 @@ namespace sw
}
else if(src.rel.type == Shader::PARAMETER_TEMP)
{
reg.x = As<Float4>(Int4(i) + As<Int4>(r.r[src.rel.index].x));
reg.x = As<Float4>(Int4(i) + As<Int4>(r[src.rel.index].x));
}
return reg;
case Shader::PARAMETER_PREDICATE: return reg; // Dummy
......@@ -812,17 +804,17 @@ namespace sw
case Shader::PARAMETER_COLOROUT:
if(src.rel.type == Shader::PARAMETER_VOID) // Not relative
{
reg = r.oC[i];
reg = oC[i];
}
else
{
Int a = relativeAddress(r, src);
Int a = relativeAddress(src);
reg = r.oC[i + a];
reg = oC[i + a];
}
break;
case Shader::PARAMETER_DEPTHOUT:
reg.x = r.oDepth;
reg.x = oDepth;
break;
default:
ASSERT(false);
......@@ -874,14 +866,14 @@ namespace sw
return mod;
}
Vector4f PixelProgram::readConstant(Registers &r, const Src &src, unsigned int offset)
Vector4f PixelProgram::readConstant(const Src &src, unsigned int offset)
{
Vector4f c;
unsigned int i = src.index + offset;
if(src.rel.type == Shader::PARAMETER_VOID) // Not relative
{
c.x = c.y = c.z = c.w = *Pointer<Float4>(r.data + OFFSET(DrawData, ps.c[i]));
c.x = c.y = c.z = c.w = *Pointer<Float4>(data + OFFSET(DrawData, ps.c[i]));
c.x = c.x.xxxx;
c.y = c.y.yyyy;
......@@ -911,9 +903,9 @@ namespace sw
}
else if(src.rel.type == Shader::PARAMETER_LOOP)
{
Int loopCounter = r.aL[r.loopDepth];
Int loopCounter = aL[loopDepth];
c.x = c.y = c.z = c.w = *Pointer<Float4>(r.data + OFFSET(DrawData, ps.c[i]) + loopCounter * 16);
c.x = c.y = c.z = c.w = *Pointer<Float4>(data + OFFSET(DrawData, ps.c[i]) + loopCounter * 16);
c.x = c.x.xxxx;
c.y = c.y.yyyy;
......@@ -922,9 +914,9 @@ namespace sw
}
else
{
Int a = relativeAddress(r, src);
Int a = relativeAddress(src);
c.x = c.y = c.z = c.w = *Pointer<Float4>(r.data + OFFSET(DrawData, ps.c[i]) + a * 16);
c.x = c.y = c.z = c.w = *Pointer<Float4>(data + OFFSET(DrawData, ps.c[i]) + a * 16);
c.x = c.x.xxxx;
c.y = c.y.yyyy;
......@@ -935,25 +927,25 @@ namespace sw
return c;
}
Int PixelProgram::relativeAddress(Registers &r, const Shader::Parameter &var)
Int PixelProgram::relativeAddress(const Shader::Parameter &var)
{
ASSERT(var.rel.deterministic);
if(var.rel.type == Shader::PARAMETER_TEMP)
{
return As<Int>(Extract(r.r[var.rel.index].x, 0)) * var.rel.scale;
return As<Int>(Extract(r[var.rel.index].x, 0)) * var.rel.scale;
}
else if(var.rel.type == Shader::PARAMETER_INPUT)
{
return As<Int>(Extract(r.v[var.rel.index].x, 0)) * var.rel.scale;
return As<Int>(Extract(v[var.rel.index].x, 0)) * var.rel.scale;
}
else if(var.rel.type == Shader::PARAMETER_OUTPUT)
{
return As<Int>(Extract(r.oC[var.rel.index].x, 0)) * var.rel.scale;
return As<Int>(Extract(oC[var.rel.index].x, 0)) * var.rel.scale;
}
else if(var.rel.type == Shader::PARAMETER_CONST)
{
RValue<Int4> c = *Pointer<Int4>(r.data + OFFSET(DrawData, ps.c[var.rel.index]));
RValue<Int4> c = *Pointer<Int4>(data + OFFSET(DrawData, ps.c[var.rel.index]));
return Extract(c, 0) * var.rel.scale;
}
......@@ -970,32 +962,32 @@ namespace sw
return Min(Max(sRGB, Float4(0.0f)), Float4(1.0f));
}
void PixelProgram::M3X2(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
void PixelProgram::M3X2(Vector4f &dst, Vector4f &src0, const Src &src1)
{
Vector4f row0 = fetchRegisterF(r, src1, 0);
Vector4f row1 = fetchRegisterF(r, src1, 1);
Vector4f row0 = fetchRegisterF(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
}
void PixelProgram::M3X3(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
void PixelProgram::M3X3(Vector4f &dst, Vector4f &src0, const Src &src1)
{
Vector4f row0 = fetchRegisterF(r, src1, 0);
Vector4f row1 = fetchRegisterF(r, src1, 1);
Vector4f row2 = fetchRegisterF(r, src1, 2);
Vector4f row0 = fetchRegisterF(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
dst.z = dot3(src0, row2);
}
void PixelProgram::M3X4(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
void PixelProgram::M3X4(Vector4f &dst, Vector4f &src0, const Src &src1)
{
Vector4f row0 = fetchRegisterF(r, src1, 0);
Vector4f row1 = fetchRegisterF(r, src1, 1);
Vector4f row2 = fetchRegisterF(r, src1, 2);
Vector4f row3 = fetchRegisterF(r, src1, 3);
Vector4f row0 = fetchRegisterF(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2);
Vector4f row3 = fetchRegisterF(src1, 3);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
......@@ -1003,23 +995,23 @@ namespace sw
dst.w = dot3(src0, row3);
}
void PixelProgram::M4X3(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
void PixelProgram::M4X3(Vector4f &dst, Vector4f &src0, const Src &src1)
{
Vector4f row0 = fetchRegisterF(r, src1, 0);
Vector4f row1 = fetchRegisterF(r, src1, 1);
Vector4f row2 = fetchRegisterF(r, src1, 2);
Vector4f row0 = fetchRegisterF(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2);
dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1);
dst.z = dot4(src0, row2);
}
void PixelProgram::M4X4(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
void PixelProgram::M4X4(Vector4f &dst, Vector4f &src0, const Src &src1)
{
Vector4f row0 = fetchRegisterF(r, src1, 0);
Vector4f row1 = fetchRegisterF(r, src1, 1);
Vector4f row2 = fetchRegisterF(r, src1, 2);
Vector4f row3 = fetchRegisterF(r, src1, 3);
Vector4f row0 = fetchRegisterF(src1, 0);
Vector4f row1 = fetchRegisterF(src1, 1);
Vector4f row2 = fetchRegisterF(src1, 2);
Vector4f row3 = fetchRegisterF(src1, 3);
dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1);
......@@ -1027,10 +1019,10 @@ namespace sw
dst.w = dot4(src0, row3);
}
void PixelProgram::TEXLD(Registers &r, 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(r, tmp, src1, src0.x, src0.y, src0.z, src0.w, src0, src0, project, bias);
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];
......@@ -1038,40 +1030,40 @@ namespace sw
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
}
void PixelProgram::TEXOFFSET(Registers &r, 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)
{
UNIMPLEMENTED();
}
void PixelProgram::TEXLDL(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &offset, bool project, bool bias)
void PixelProgram::TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &offset, bool project, bool bias)
{
UNIMPLEMENTED();
}
void PixelProgram::TEXELFETCH(Registers &r, Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2)
void PixelProgram::TEXELFETCH(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2)
{
UNIMPLEMENTED();
}
void PixelProgram::TEXELFETCH(Registers &r, Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &offset)
void PixelProgram::TEXELFETCH(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &offset)
{
UNIMPLEMENTED();
}
void PixelProgram::TEXGRAD(Registers &r, Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3)
void PixelProgram::TEXGRAD(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3)
{
UNIMPLEMENTED();
}
void PixelProgram::TEXGRAD(Registers &r, Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3, Vector4f &offset)
void PixelProgram::TEXGRAD(Vector4f &dst, Vector4f &src0, const Src& src1, Vector4f &src2, Vector4f &src3, Vector4f &offset)
{
UNIMPLEMENTED();
}
void PixelProgram::TEXLDD(Registers &r, 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, bool bias)
{
Vector4f tmp;
sampleTexture(r, tmp, src1, src0.x, src0.y, src0.z, src0.w, src2, src3, project, bias, true);
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];
......@@ -1079,10 +1071,10 @@ namespace sw
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
}
void PixelProgram::TEXLDL(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias)
void PixelProgram::TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias)
{
Vector4f tmp;
sampleTexture(r, tmp, src1, src0.x, src0.y, src0.z, src0.w, src0, src0, project, bias, false, true);
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];
......@@ -1090,9 +1082,9 @@ namespace sw
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
}
void PixelProgram::TEXSIZE(Registers &r, Vector4f &dst, Float4 &lod, const Src &src1)
void PixelProgram::TEXSIZE(Vector4f &dst, Float4 &lod, const Src &src1)
{
Pointer<Byte> textureMipmap = r.data + OFFSET(DrawData, mipmap) + src1.index * sizeof(Texture) + OFFSET(Texture, mipmap);
Pointer<Byte> textureMipmap = data + OFFSET(DrawData, mipmap) + src1.index * sizeof(Texture) + OFFSET(Texture, mipmap);
for(int i = 0; i < 4; ++i)
{
Pointer<Byte> mipmap = textureMipmap + (As<Int>(Extract(lod, i)) + Int(1)) * sizeof(Mipmap);
......@@ -1114,7 +1106,7 @@ namespace sw
// FIXME: Dynamic branching affects TEXKILL?
// if(shader->containsDynamicBranching())
// {
// kill = ~SignMask(enableMask(r));
// kill = ~SignMask(enableMask());
// }
for(unsigned int q = 0; q < state.multiSample; q++)
......@@ -1125,13 +1117,13 @@ namespace sw
// FIXME: Branch to end of shader if all killed?
}
void PixelProgram::DISCARD(Registers &r, Int cMask[4], const Shader::Instruction *instruction)
void PixelProgram::DISCARD(Int cMask[4], const Shader::Instruction *instruction)
{
Int kill = 0;
if(shader->containsDynamicBranching())
{
kill = ~SignMask(enableMask(r, instruction));
kill = ~SignMask(enableMask(instruction));
}
for(unsigned int q = 0; q < state.multiSample; q++)
......@@ -1167,30 +1159,30 @@ namespace sw
dst.w = Abs(src.w.yyww - src.w.xxzz) + Abs(src.w.zwzw - src.w.xyxy);
}
void PixelProgram::BREAK(Registers &r)
void PixelProgram::BREAK()
{
llvm::BasicBlock *deadBlock = Nucleus::createBasicBlock();
llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth - 1];
if(breakDepth == 0)
{
r.enableIndex = r.enableIndex - breakDepth;
enableIndex = enableIndex - breakDepth;
Nucleus::createBr(endBlock);
}
else
{
r.enableBreak = r.enableBreak & ~r.enableStack[r.enableIndex];
Bool allBreak = SignMask(r.enableBreak) == 0x0;
enableBreak = enableBreak & ~enableStack[enableIndex];
Bool allBreak = SignMask(enableBreak) == 0x0;
r.enableIndex = r.enableIndex - breakDepth;
enableIndex = enableIndex - breakDepth;
branch(allBreak, endBlock, deadBlock);
}
Nucleus::setInsertBlock(deadBlock);
r.enableIndex = r.enableIndex + breakDepth;
enableIndex = enableIndex + breakDepth;
}
void PixelProgram::BREAKC(Registers &r, Vector4f &src0, Vector4f &src1, Control control)
void PixelProgram::BREAKC(Vector4f &src0, Vector4f &src1, Control control)
{
Int4 condition;
......@@ -1206,41 +1198,41 @@ namespace sw
ASSERT(false);
}
BREAK(r, condition);
BREAK(condition);
}
void PixelProgram::BREAKP(Registers &r, const Src &predicateRegister) // FIXME: Factor out parts common with BREAKC
void PixelProgram::BREAKP(const Src &predicateRegister) // FIXME: Factor out parts common with BREAKC
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
Int4 condition = As<Int4>(p0[predicateRegister.swizzle & 0x3]);
if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
BREAK(r, condition);
BREAK(condition);
}
void PixelProgram::BREAK(Registers &r, Int4 &condition)
void PixelProgram::BREAK(Int4 &condition)
{
condition &= r.enableStack[r.enableIndex];
condition &= enableStack[enableIndex];
llvm::BasicBlock *continueBlock = Nucleus::createBasicBlock();
llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth - 1];
r.enableBreak = r.enableBreak & ~condition;
Bool allBreak = SignMask(r.enableBreak) == 0x0;
enableBreak = enableBreak & ~condition;
Bool allBreak = SignMask(enableBreak) == 0x0;
r.enableIndex = r.enableIndex - breakDepth;
enableIndex = enableIndex - breakDepth;
branch(allBreak, endBlock, continueBlock);
Nucleus::setInsertBlock(continueBlock);
r.enableIndex = r.enableIndex + breakDepth;
enableIndex = enableIndex + breakDepth;
}
void PixelProgram::CONTINUE(Registers &r)
void PixelProgram::CONTINUE()
{
r.enableContinue = r.enableContinue & ~r.enableStack[r.enableIndex];
enableContinue = enableContinue & ~enableStack[enableIndex];
}
void PixelProgram::TEST()
......@@ -1248,7 +1240,7 @@ namespace sw
whileTest = true;
}
void PixelProgram::CALL(Registers &r, int labelIndex, int callSiteIndex)
void PixelProgram::CALL(int labelIndex, int callSiteIndex)
{
if(!labelBlock[labelIndex])
{
......@@ -1257,33 +1249,33 @@ namespace sw
if(callRetBlock[labelIndex].size() > 1)
{
r.callStack[r.stackIndex++] = UInt(callSiteIndex);
callStack[stackIndex++] = UInt(callSiteIndex);
}
Int4 restoreLeave = r.enableLeave;
Int4 restoreLeave = enableLeave;
Nucleus::createBr(labelBlock[labelIndex]);
Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
r.enableLeave = restoreLeave;
enableLeave = restoreLeave;
}
void PixelProgram::CALLNZ(Registers &r, int labelIndex, int callSiteIndex, const Src &src)
void PixelProgram::CALLNZ(int labelIndex, int callSiteIndex, const Src &src)
{
if(src.type == Shader::PARAMETER_CONSTBOOL)
{
CALLNZb(r, labelIndex, callSiteIndex, src);
CALLNZb(labelIndex, callSiteIndex, src);
}
else if(src.type == Shader::PARAMETER_PREDICATE)
{
CALLNZp(r, labelIndex, callSiteIndex, src);
CALLNZp(labelIndex, callSiteIndex, src);
}
else ASSERT(false);
}
void PixelProgram::CALLNZb(Registers &r, int labelIndex, int callSiteIndex, const Src &boolRegister)
void PixelProgram::CALLNZb(int labelIndex, int callSiteIndex, const Src &boolRegister)
{
Bool condition = (*Pointer<Byte>(r.data + OFFSET(DrawData, ps.b[boolRegister.index])) != Byte(0)); // FIXME
Bool condition = (*Pointer<Byte>(data + OFFSET(DrawData, ps.b[boolRegister.index])) != Byte(0)); // FIXME
if(boolRegister.modifier == Shader::MODIFIER_NOT)
{
......@@ -1297,27 +1289,27 @@ namespace sw
if(callRetBlock[labelIndex].size() > 1)
{
r.callStack[r.stackIndex++] = UInt(callSiteIndex);
callStack[stackIndex++] = UInt(callSiteIndex);
}
Int4 restoreLeave = r.enableLeave;
Int4 restoreLeave = enableLeave;
branch(condition, labelBlock[labelIndex], callRetBlock[labelIndex][callSiteIndex]);
Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
r.enableLeave = restoreLeave;
enableLeave = restoreLeave;
}
void PixelProgram::CALLNZp(Registers &r, int labelIndex, int callSiteIndex, const Src &predicateRegister)
void PixelProgram::CALLNZp(int labelIndex, int callSiteIndex, const Src &predicateRegister)
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
Int4 condition = As<Int4>(p0[predicateRegister.swizzle & 0x3]);
if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
condition &= r.enableStack[r.enableIndex];
condition &= enableStack[enableIndex];
if(!labelBlock[labelIndex])
{
......@@ -1326,22 +1318,22 @@ namespace sw
if(callRetBlock[labelIndex].size() > 1)
{
r.callStack[r.stackIndex++] = UInt(callSiteIndex);
callStack[stackIndex++] = UInt(callSiteIndex);
}
r.enableIndex++;
r.enableStack[r.enableIndex] = condition;
Int4 restoreLeave = r.enableLeave;
enableIndex++;
enableStack[enableIndex] = condition;
Int4 restoreLeave = enableLeave;
Bool notAllFalse = SignMask(condition) != 0;
branch(notAllFalse, labelBlock[labelIndex], callRetBlock[labelIndex][callSiteIndex]);
Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
r.enableIndex--;
r.enableLeave = restoreLeave;
enableIndex--;
enableLeave = restoreLeave;
}
void PixelProgram::ELSE(Registers &r)
void PixelProgram::ELSE()
{
ifDepth--;
......@@ -1350,12 +1342,12 @@ namespace sw
if(isConditionalIf[ifDepth])
{
Int4 condition = ~r.enableStack[r.enableIndex] & r.enableStack[r.enableIndex - 1];
Int4 condition = ~enableStack[enableIndex] & enableStack[enableIndex - 1];
Bool notAllFalse = SignMask(condition) != 0;
branch(notAllFalse, falseBlock, endBlock);
r.enableStack[r.enableIndex] = ~r.enableStack[r.enableIndex] & r.enableStack[r.enableIndex - 1];
enableStack[enableIndex] = ~enableStack[enableIndex] & enableStack[enableIndex - 1];
}
else
{
......@@ -1368,7 +1360,7 @@ namespace sw
ifDepth++;
}
void PixelProgram::ENDIF(Registers &r)
void PixelProgram::ENDIF()
{
ifDepth--;
......@@ -1380,15 +1372,15 @@ namespace sw
if(isConditionalIf[ifDepth])
{
breakDepth--;
r.enableIndex--;
enableIndex--;
}
}
void PixelProgram::ENDLOOP(Registers &r)
void PixelProgram::ENDLOOP()
{
loopRepDepth--;
r.aL[r.loopDepth] = r.aL[r.loopDepth] + r.increment[r.loopDepth]; // FIXME: +=
aL[loopDepth] = aL[loopDepth] + increment[loopDepth]; // FIXME: +=
llvm::BasicBlock *testBlock = loopRepTestBlock[loopRepDepth];
llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth];
......@@ -1396,11 +1388,11 @@ namespace sw
Nucleus::createBr(testBlock);
Nucleus::setInsertBlock(endBlock);
r.loopDepth--;
r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
loopDepth--;
enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
void PixelProgram::ENDREP(Registers &r)
void PixelProgram::ENDREP()
{
loopRepDepth--;
......@@ -1410,11 +1402,11 @@ namespace sw
Nucleus::createBr(testBlock);
Nucleus::setInsertBlock(endBlock);
r.loopDepth--;
r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
loopDepth--;
enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
void PixelProgram::ENDWHILE(Registers &r)
void PixelProgram::ENDWHILE()
{
loopRepDepth--;
......@@ -1424,33 +1416,33 @@ namespace sw
Nucleus::createBr(testBlock);
Nucleus::setInsertBlock(endBlock);
r.enableIndex--;
r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
enableIndex--;
enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
whileTest = false;
}
void PixelProgram::IF(Registers &r, const Src &src)
void PixelProgram::IF(const Src &src)
{
if(src.type == Shader::PARAMETER_CONSTBOOL)
{
IFb(r, src);
IFb(src);
}
else if(src.type == Shader::PARAMETER_PREDICATE)
{
IFp(r, src);
IFp(src);
}
else
{
Int4 condition = As<Int4>(fetchRegisterF(r, src).x);
IF(r, condition);
Int4 condition = As<Int4>(fetchRegisterF(src).x);
IF(condition);
}
}
void PixelProgram::IFb(Registers &r, const Src &boolRegister)
void PixelProgram::IFb(const Src &boolRegister)
{
ASSERT(ifDepth < 24 + 4);
Bool condition = (*Pointer<Byte>(r.data + OFFSET(DrawData, ps.b[boolRegister.index])) != Byte(0)); // FIXME
Bool condition = (*Pointer<Byte>(data + OFFSET(DrawData, ps.b[boolRegister.index])) != Byte(0)); // FIXME
if(boolRegister.modifier == Shader::MODIFIER_NOT)
{
......@@ -1468,19 +1460,19 @@ namespace sw
ifDepth++;
}
void PixelProgram::IFp(Registers &r, const Src &predicateRegister)
void PixelProgram::IFp(const Src &predicateRegister)
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
Int4 condition = As<Int4>(p0[predicateRegister.swizzle & 0x3]);
if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
IF(r, condition);
IF(condition);
}
void PixelProgram::IFC(Registers &r, Vector4f &src0, Vector4f &src1, Control control)
void PixelProgram::IFC(Vector4f &src0, Vector4f &src1, Control control)
{
Int4 condition;
......@@ -1496,15 +1488,15 @@ namespace sw
ASSERT(false);
}
IF(r, condition);
IF(condition);
}
void PixelProgram::IF(Registers &r, Int4 &condition)
void PixelProgram::IF(Int4 &condition)
{
condition &= r.enableStack[r.enableIndex];
condition &= enableStack[enableIndex];
r.enableIndex++;
r.enableStack[r.enableIndex] = condition;
enableIndex++;
enableStack[enableIndex] = condition;
llvm::BasicBlock *trueBlock = Nucleus::createBasicBlock();
llvm::BasicBlock *falseBlock = Nucleus::createBasicBlock();
......@@ -1531,17 +1523,17 @@ namespace sw
currentLabel = labelIndex;
}
void PixelProgram::LOOP(Registers &r, const Src &integerRegister)
void PixelProgram::LOOP(const Src &integerRegister)
{
r.loopDepth++;
loopDepth++;
r.iteration[r.loopDepth] = *Pointer<Int>(r.data + OFFSET(DrawData, ps.i[integerRegister.index][0]));
r.aL[r.loopDepth] = *Pointer<Int>(r.data + OFFSET(DrawData, ps.i[integerRegister.index][1]));
r.increment[r.loopDepth] = *Pointer<Int>(r.data + OFFSET(DrawData, ps.i[integerRegister.index][2]));
iteration[loopDepth] = *Pointer<Int>(data + OFFSET(DrawData, ps.i[integerRegister.index][0]));
aL[loopDepth] = *Pointer<Int>(data + OFFSET(DrawData, ps.i[integerRegister.index][1]));
increment[loopDepth] = *Pointer<Int>(data + OFFSET(DrawData, ps.i[integerRegister.index][2]));
// If(r.increment[r.loopDepth] == 0)
// If(increment[loopDepth] == 0)
// {
// r.increment[r.loopDepth] = 1;
// increment[loopDepth] = 1;
// }
llvm::BasicBlock *loopBlock = Nucleus::createBasicBlock();
......@@ -1555,21 +1547,21 @@ namespace sw
Nucleus::createBr(testBlock);
Nucleus::setInsertBlock(testBlock);
branch(r.iteration[r.loopDepth] > 0, loopBlock, endBlock);
branch(iteration[loopDepth] > 0, loopBlock, endBlock);
Nucleus::setInsertBlock(loopBlock);
r.iteration[r.loopDepth] = r.iteration[r.loopDepth] - 1; // FIXME: --
iteration[loopDepth] = iteration[loopDepth] - 1; // FIXME: --
loopRepDepth++;
breakDepth = 0;
}
void PixelProgram::REP(Registers &r, const Src &integerRegister)
void PixelProgram::REP(const Src &integerRegister)
{
r.loopDepth++;
loopDepth++;
r.iteration[r.loopDepth] = *Pointer<Int>(r.data + OFFSET(DrawData, ps.i[integerRegister.index][0]));
r.aL[r.loopDepth] = r.aL[r.loopDepth - 1];
iteration[loopDepth] = *Pointer<Int>(data + OFFSET(DrawData, ps.i[integerRegister.index][0]));
aL[loopDepth] = aL[loopDepth - 1];
llvm::BasicBlock *loopBlock = Nucleus::createBasicBlock();
llvm::BasicBlock *testBlock = Nucleus::createBasicBlock();
......@@ -1582,18 +1574,18 @@ namespace sw
Nucleus::createBr(testBlock);
Nucleus::setInsertBlock(testBlock);
branch(r.iteration[r.loopDepth] > 0, loopBlock, endBlock);
branch(iteration[loopDepth] > 0, loopBlock, endBlock);
Nucleus::setInsertBlock(loopBlock);
r.iteration[r.loopDepth] = r.iteration[r.loopDepth] - 1; // FIXME: --
iteration[loopDepth] = iteration[loopDepth] - 1; // FIXME: --
loopRepDepth++;
breakDepth = 0;
}
void PixelProgram::WHILE(Registers &r, const Src &temporaryRegister)
void PixelProgram::WHILE(const Src &temporaryRegister)
{
r.enableIndex++;
enableIndex++;
llvm::BasicBlock *loopBlock = Nucleus::createBasicBlock();
llvm::BasicBlock *testBlock = Nucleus::createBasicBlock();
......@@ -1602,24 +1594,24 @@ namespace sw
loopRepTestBlock[loopRepDepth] = testBlock;
loopRepEndBlock[loopRepDepth] = endBlock;
Int4 restoreBreak = r.enableBreak;
Int4 restoreContinue = r.enableContinue;
Int4 restoreBreak = enableBreak;
Int4 restoreContinue = enableContinue;
// FIXME: jump(testBlock)
Nucleus::createBr(testBlock);
Nucleus::setInsertBlock(testBlock);
r.enableContinue = restoreContinue;
enableContinue = restoreContinue;
const Vector4f &src = fetchRegisterF(r, temporaryRegister);
const Vector4f &src = fetchRegisterF(temporaryRegister);
Int4 condition = As<Int4>(src.x);
condition &= r.enableStack[r.enableIndex - 1];
r.enableStack[r.enableIndex] = condition;
condition &= enableStack[enableIndex - 1];
enableStack[enableIndex] = condition;
Bool notAllFalse = SignMask(condition) != 0;
branch(notAllFalse, loopBlock, endBlock);
Nucleus::setInsertBlock(endBlock);
r.enableBreak = restoreBreak;
enableBreak = restoreBreak;
Nucleus::setInsertBlock(loopBlock);
......@@ -1627,7 +1619,7 @@ namespace sw
breakDepth = 0;
}
void PixelProgram::RET(Registers &r)
void PixelProgram::RET()
{
if(currentLabel == -1)
{
......@@ -1641,7 +1633,7 @@ namespace sw
if(callRetBlock[currentLabel].size() > 1) // Pop the return destination from the call stack
{
// FIXME: Encapsulate
UInt index = r.callStack[--r.stackIndex];
UInt index = callStack[--stackIndex];
llvm::Value *value = index.loadValue();
llvm::Value *switchInst = Nucleus::createSwitch(value, unreachableBlock, (int)callRetBlock[currentLabel].size());
......@@ -1665,9 +1657,9 @@ namespace sw
}
}
void PixelProgram::LEAVE(Registers &r)
void PixelProgram::LEAVE()
{
r.enableLeave = r.enableLeave & ~r.enableStack[r.enableIndex];
enableLeave = enableLeave & ~enableStack[enableIndex];
// FIXME: Return from function if all instances left
// FIXME: Use enableLeave in other control-flow constructs
......
src/Shader/PixelProgram.hpp
......@@ -20,148 +20,141 @@ namespace sw
{
public:
PixelProgram(const PixelProcessor::State &state, const PixelShader *shader) :
PixelRoutine(state, shader), ifDepth(0), loopRepDepth(0), breakDepth(0), currentLabel(-1), whileTest(false)
PixelRoutine(state, shader), r(shader && shader->dynamicallyIndexedTemporaries),
loopDepth(-1), ifDepth(0), loopRepDepth(0), breakDepth(0), currentLabel(-1), whileTest(false)
{
for(int i = 0; i < 2048; ++i)
{
labelBlock[i] = 0;
for(int i = 0; i < 2048; ++i)
{
labelBlock[i] = 0;
}
enableStack[0] = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
if(shader && shader->containsBreakInstruction())
{
enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
if(shader && shader->containsContinueInstruction())
{
enableContinue = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
}
virtual ~PixelProgram() {}
protected:
virtual void setBuiltins(PixelRoutine::Registers &r, Int &x, Int &y, Float4(&z)[4], Float4 &w);
virtual void applyShader(PixelRoutine::Registers &r, Int cMask[4]);
virtual Bool alphaTest(PixelRoutine::Registers &r, Int cMask[4]);
virtual void rasterOperation(PixelRoutine::Registers &r, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
virtual QuadRasterizer::Registers* createRegisters(const PixelShader *shader) { return new PixelProgram::Registers(shader); };
virtual void setBuiltins(Int &x, Int &y, Float4(&z)[4], Float4 &w);
virtual void applyShader(Int cMask[4]);
virtual Bool alphaTest(Int cMask[4]);
virtual void rasterOperation(Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
private:
struct Registers : public PixelRoutine::Registers
{
Registers(const PixelShader *shader) :
PixelRoutine::Registers(shader),
r(shader && shader->dynamicallyIndexedTemporaries),
loopDepth(-1)
{
enableStack[0] = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
if(shader && shader->containsBreakInstruction())
{
enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
if(shader && shader->containsContinueInstruction())
{
enableContinue = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
}
// Temporary registers
RegisterArray<4096> r;
// Color outputs
Vector4f c[RENDERTARGETS];
RegisterArray<RENDERTARGETS, true> oC;
// Shader variables
Vector4f vPos;
Vector4f vFace;
// DX9 specific variables
Vector4f p0;
Array<Int, 4> aL;
Array<Int, 4> increment;
Array<Int, 4> iteration;
Int loopDepth; // FIXME: Add support for switch
Int stackIndex; // FIXME: Inc/decrement callStack
Array<UInt, 16> callStack;
// Per pixel based on conditions reached
Int enableIndex;
Array<Int4, 1 + 24> enableStack;
Int4 enableBreak;
Int4 enableContinue;
Int4 enableLeave;
};
void sampleTexture(Registers &r, 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(Registers &r, 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);
// Raster operations
void clampColor(Vector4f oC[RENDERTARGETS]);
Int4 enableMask(Registers &r, const Shader::Instruction *instruction);
Vector4f fetchRegisterF(Registers &r, const Src &src, unsigned int offset = 0);
Vector4f readConstant(Registers &r, const Src &src, unsigned int offset = 0);
Int relativeAddress(Registers &r, const Shader::Parameter &var);
Float4 linearToSRGB(const Float4 &x);
// Instructions
typedef Shader::Control Control;
void M3X2(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1);
void M3X3(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1);
void M3X4(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1);
void M4X3(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1);
void M4X4(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1);
void TEXLD(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias);
void TEXLDD(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, Vector4f &src3, bool project, bool bias);
void TEXLDL(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias);
void TEXSIZE(Registers &r, Vector4f &dst, Float4 &lod, const Src &src1);
void TEXKILL(Int cMask[4], Vector4f &src, unsigned char mask);
void TEXOFFSET(Registers &r, Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3, bool project, bool bias);
void TEXLDL(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, bool project, bool bias);
void TEXELFETCH(Registers &r, Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2);
void TEXELFETCH(Registers &r, Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3);
void TEXGRAD(Registers &r, Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3);
void TEXGRAD(Registers &r, Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3, Vector4f &src4);
void DISCARD(Registers &r, Int cMask[4], const Shader::Instruction *instruction);
void DFDX(Vector4f &dst, Vector4f &src);
void DFDY(Vector4f &dst, Vector4f &src);
void FWIDTH(Vector4f &dst, Vector4f &src);
void BREAK(Registers &r);
void BREAKC(Registers &r, Vector4f &src0, Vector4f &src1, Control);
void BREAKP(Registers &r, const Src &predicateRegister);
void BREAK(Registers &r, Int4 &condition);
void CONTINUE(Registers &r);
void TEST();
void CALL(Registers &r, int labelIndex, int callSiteIndex);
void CALLNZ(Registers &r, int labelIndex, int callSiteIndex, const Src &src);
void CALLNZb(Registers &r, int labelIndex, int callSiteIndex, const Src &boolRegister);
void CALLNZp(Registers &r, int labelIndex, int callSiteIndex, const Src &predicateRegister);
void ELSE(Registers &r);
void ENDIF(Registers &r);
void ENDLOOP(Registers &r);
void ENDREP(Registers &r);
void ENDWHILE(Registers &r);
void IF(Registers &r, const Src &src);
void IFb(Registers &r, const Src &boolRegister);
void IFp(Registers &r, const Src &predicateRegister);
void IFC(Registers &r, Vector4f &src0, Vector4f &src1, Control);
void IF(Registers &r, Int4 &condition);
void LABEL(int labelIndex);
void LOOP(Registers &r, const Src &integerRegister);
void REP(Registers &r, const Src &integerRegister);
void WHILE(Registers &r, const Src &temporaryRegister);
void RET(Registers &r);
void LEAVE(Registers &r);
int ifDepth;
int loopRepDepth;
int breakDepth;
int currentLabel;
// Temporary registers
RegisterArray<4096> r;
// Color outputs
Vector4f c[RENDERTARGETS];
RegisterArray<RENDERTARGETS, true> oC;
// Shader variables
Vector4f vPos;
Vector4f vFace;
// DX9 specific variables
Vector4f p0;
Array<Int, 4> aL;
Array<Int, 4> increment;
Array<Int, 4> iteration;
Int loopDepth; // FIXME: Add support for switch
Int stackIndex; // FIXME: Inc/decrement callStack
Array<UInt, 16> callStack;
// Per pixel based on conditions reached
Int enableIndex;
Array<Int4, 1 + 24> enableStack;
Int4 enableBreak;
Int4 enableContinue;
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, 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);
// Raster operations
void clampColor(Vector4f oC[RENDERTARGETS]);
Int4 enableMask(const Shader::Instruction *instruction);
Vector4f fetchRegisterF(const Src &src, unsigned int offset = 0);
Vector4f readConstant(const Src &src, unsigned int offset = 0);
Int relativeAddress(const Shader::Parameter &var);
Float4 linearToSRGB(const Float4 &x);
// Instructions
typedef Shader::Control Control;
void M3X2(Vector4f &dst, Vector4f &src0, const Src &src1);
void M3X3(Vector4f &dst, Vector4f &src0, const Src &src1);
void M3X4(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 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 TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias);
void TEXSIZE(Vector4f &dst, Float4 &lod, const Src &src1);
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 TEXLDL(Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, bool project, bool bias);
void TEXELFETCH(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2);
void TEXELFETCH(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3);
void TEXGRAD(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3);
void TEXGRAD(Vector4f &dst, Vector4f &src, const Src&, Vector4f &src2, Vector4f &src3, Vector4f &src4);
void DISCARD(Int cMask[4], const Shader::Instruction *instruction);
void DFDX(Vector4f &dst, Vector4f &src);
void DFDY(Vector4f &dst, Vector4f &src);
void FWIDTH(Vector4f &dst, Vector4f &src);
void BREAK();
void BREAKC(Vector4f &src0, Vector4f &src1, Control);
void BREAKP(const Src &predicateRegister);
void BREAK(Int4 &condition);
void CONTINUE();
void TEST();
void CALL(int labelIndex, int callSiteIndex);
void CALLNZ(int labelIndex, int callSiteIndex, const Src &src);
void CALLNZb(int labelIndex, int callSiteIndex, const Src &boolRegister);
void CALLNZp(int labelIndex, int callSiteIndex, const Src &predicateRegister);
void ELSE();
void ENDIF();
void ENDLOOP();
void ENDREP();
void ENDWHILE();
void IF(const Src &src);
void IFb(const Src &boolRegister);
void IFp(const Src &predicateRegister);
void IFC(Vector4f &src0, Vector4f &src1, Control);
void IF(Int4 &condition);
void LABEL(int labelIndex);
void LOOP(const Src &integerRegister);
void REP(const Src &integerRegister);
void WHILE(const Src &temporaryRegister);
void RET();
void LEAVE();
int ifDepth;
int loopRepDepth;
int breakDepth;
int currentLabel;
bool whileTest;
// FIXME: Get rid of llvm::
llvm::BasicBlock *ifFalseBlock[24 + 24];
llvm::BasicBlock *loopRepTestBlock[4];
llvm::BasicBlock *loopRepEndBlock[4];
llvm::BasicBlock *labelBlock[2048];
std::vector<llvm::BasicBlock*> callRetBlock[2048];
llvm::BasicBlock *returnBlock;
// FIXME: Get rid of llvm::
llvm::BasicBlock *ifFalseBlock[24 + 24];
llvm::BasicBlock *loopRepTestBlock[4];
llvm::BasicBlock *loopRepEndBlock[4];
llvm::BasicBlock *labelBlock[2048];
std::vector<llvm::BasicBlock*> callRetBlock[2048];
llvm::BasicBlock *returnBlock;
bool isConditionalIf[24 + 24];
};
}
......
src/Shader/PixelRoutine.cpp
......@@ -27,9 +27,7 @@ namespace sw
extern bool exactColorRounding;
extern bool forceClearRegisters;
PixelRoutine::Registers::Registers(const PixelShader *shader) :
QuadRasterizer::Registers(),
v(shader && shader->dynamicallyIndexedInput)
PixelRoutine::PixelRoutine(const PixelProcessor::State &state, const PixelShader *shader) : QuadRasterizer(state, shader), v(shader && shader->dynamicallyIndexedInput)
{
if(!shader || shader->getVersion() < 0x0200 || forceClearRegisters)
{
......@@ -43,10 +41,6 @@ namespace sw
}
}
PixelRoutine::PixelRoutine(const PixelProcessor::State &state, const PixelShader *shader) : QuadRasterizer(state, shader)
{
}
PixelRoutine::~PixelRoutine()
{
for(int i = 0; i < TEXTURE_IMAGE_UNITS; i++)
......@@ -55,17 +49,15 @@ namespace sw
}
}
void PixelRoutine::quad(QuadRasterizer::Registers &rBase, Pointer<Byte> cBuffer[RENDERTARGETS], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y)
void PixelRoutine::quad(Pointer<Byte> cBuffer[RENDERTARGETS], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y)
{
Registers& r = *static_cast<Registers*>(&rBase);
#if PERF_PROFILE
Long pipeTime = Ticks();
#endif
for(int i = 0; i < TEXTURE_IMAGE_UNITS; i++)
{
sampler[i] = new SamplerCore(r.constants, state.sampler[i]);
sampler[i] = new SamplerCore(constants, state.sampler[i]);
}
const bool earlyDepthTest = !state.depthOverride && !state.alphaTestActive();
......@@ -81,30 +73,26 @@ namespace sw
for(unsigned int q = 0; q < state.multiSample; q++)
{
stencilTest(r, sBuffer, q, x, sMask[q], cMask[q]);
stencilTest(sBuffer, q, x, sMask[q], cMask[q]);
}
Float4 f;
Float4 (&z)[4] = r.z;
Float4 &w = r.w;
Float4 &rhw = r.rhw;
Float4 rhwCentroid;
Float4 xxxx = Float4(Float(x)) + *Pointer<Float4>(r.primitive + OFFSET(Primitive,xQuad), 16);
Float4 xxxx = Float4(Float(x)) + *Pointer<Float4>(primitive + OFFSET(Primitive,xQuad), 16);
if(interpolateZ())
{
for(unsigned int q = 0; q < state.multiSample; q++)
{
Float4 x = xxxx;
if(state.multiSample > 1)
{
x -= *Pointer<Float4>(r.constants + OFFSET(Constants,X) + q * sizeof(float4));
x -= *Pointer<Float4>(constants + OFFSET(Constants,X) + q * sizeof(float4));
}
z[q] = interpolate(x, r.Dz[q], z[q], r.primitive + OFFSET(Primitive,z), false, false);
z[q] = interpolate(x, Dz[q], z[q], primitive + OFFSET(Primitive,z), false, false);
}
}
......@@ -114,7 +102,7 @@ namespace sw
{
for(unsigned int q = 0; q < state.multiSample; q++)
{
depthPass = depthPass || depthTest(r, zBuffer, q, x, z[q], sMask[q], zMask[q], cMask[q]);
depthPass = depthPass || depthTest(zBuffer, q, x, z[q], sMask[q], zMask[q], cMask[q]);
}
}
......@@ -124,7 +112,7 @@ namespace sw
Long interpTime = Ticks();
#endif
Float4 yyyy = Float4(Float(y)) + *Pointer<Float4>(r.primitive + OFFSET(Primitive,yQuad), 16);
Float4 yyyy = Float4(Float(y)) + *Pointer<Float4>(primitive + OFFSET(Primitive,yQuad), 16);
// Centroid locations
Float4 XXXX = Float4(0.0f);
......@@ -136,9 +124,9 @@ namespace sw
for(unsigned int q = 0; q < state.multiSample; q++)
{
XXXX += *Pointer<Float4>(r.constants + OFFSET(Constants,sampleX[q]) + 16 * cMask[q]);
YYYY += *Pointer<Float4>(r.constants + OFFSET(Constants,sampleY[q]) + 16 * cMask[q]);
WWWW += *Pointer<Float4>(r.constants + OFFSET(Constants,weight) + 16 * cMask[q]);
XXXX += *Pointer<Float4>(constants + OFFSET(Constants,sampleX[q]) + 16 * cMask[q]);
YYYY += *Pointer<Float4>(constants + OFFSET(Constants,sampleY[q]) + 16 * cMask[q]);
WWWW += *Pointer<Float4>(constants + OFFSET(Constants,weight) + 16 * cMask[q]);
}
WWWW = Rcp_pp(WWWW);
......@@ -151,12 +139,12 @@ namespace sw
if(interpolateW())
{
w = interpolate(xxxx, r.Dw, rhw, r.primitive + OFFSET(Primitive,w), false, false);
w = interpolate(xxxx, Dw, rhw, primitive + OFFSET(Primitive,w), false, false);
rhw = reciprocal(w);
if(state.centroid)
{
rhwCentroid = reciprocal(interpolateCentroid(XXXX, YYYY, rhwCentroid, r.primitive + OFFSET(Primitive,w), false, false));
rhwCentroid = reciprocal(interpolateCentroid(XXXX, YYYY, rhwCentroid, primitive + OFFSET(Primitive,w), false, false));
}
}
......@@ -168,11 +156,11 @@ namespace sw
{
if(!state.interpolant[interpolant].centroid)
{
r.v[interpolant][component] = interpolate(xxxx, r.Dv[interpolant][component], rhw, r.primitive + OFFSET(Primitive, V[interpolant][component]), (state.interpolant[interpolant].flat & (1 << component)) != 0, state.perspective);
v[interpolant][component] = interpolate(xxxx, Dv[interpolant][component], rhw, primitive + OFFSET(Primitive, V[interpolant][component]), (state.interpolant[interpolant].flat & (1 << component)) != 0, state.perspective);
}
else
{
r.v[interpolant][component] = interpolateCentroid(XXXX, YYYY, rhwCentroid, r.primitive + OFFSET(Primitive, V[interpolant][component]), (state.interpolant[interpolant].flat & (1 << component)) != 0, state.perspective);
v[interpolant][component] = interpolateCentroid(XXXX, YYYY, rhwCentroid, primitive + OFFSET(Primitive, V[interpolant][component]), (state.interpolant[interpolant].flat & (1 << component)) != 0, state.perspective);
}
}
}
......@@ -184,32 +172,32 @@ namespace sw
case 0:
break;
case 1:
rcp = reciprocal(r.v[interpolant].y);
r.v[interpolant].x = r.v[interpolant].x * rcp;
rcp = reciprocal(v[interpolant].y);
v[interpolant].x = v[interpolant].x * rcp;
break;
case 2:
rcp = reciprocal(r.v[interpolant].z);
r.v[interpolant].x = r.v[interpolant].x * rcp;
r.v[interpolant].y = r.v[interpolant].y * rcp;
rcp = reciprocal(v[interpolant].z);
v[interpolant].x = v[interpolant].x * rcp;
v[interpolant].y = v[interpolant].y * rcp;
break;
case 3:
rcp = reciprocal(r.v[interpolant].w);
r.v[interpolant].x = r.v[interpolant].x * rcp;
r.v[interpolant].y = r.v[interpolant].y * rcp;
r.v[interpolant].z = r.v[interpolant].z * rcp;
rcp = reciprocal(v[interpolant].w);
v[interpolant].x = v[interpolant].x * rcp;
v[interpolant].y = v[interpolant].y * rcp;
v[interpolant].z = v[interpolant].z * rcp;
break;
}
}
if(state.fog.component)
{
f = interpolate(xxxx, r.Df, rhw, r.primitive + OFFSET(Primitive,f), state.fog.flat & 0x01, state.perspective);
f = interpolate(xxxx, Df, rhw, primitive + OFFSET(Primitive,f), state.fog.flat & 0x01, state.perspective);
}
setBuiltins(r, x, y, z, w);
setBuiltins(x, y, z, w);
#if PERF_PROFILE
r.cycles[PERF_INTERP] += Ticks() - interpTime;
cycles[PERF_INTERP] += Ticks() - interpTime;
#endif
Bool alphaPass = true;
......@@ -220,13 +208,13 @@ namespace sw
Long shaderTime = Ticks();
#endif
applyShader(r, cMask);
applyShader(cMask);
#if PERF_PROFILE
r.cycles[PERF_SHADER] += Ticks() - shaderTime;
cycles[PERF_SHADER] += Ticks() - shaderTime;
#endif
alphaPass = alphaTest(r, cMask);
alphaPass = alphaTest(cMask);
if((shader && shader->containsKill()) || state.alphaTestActive())
{
......@@ -244,7 +232,7 @@ namespace sw
{
for(unsigned int q = 0; q < state.multiSample; q++)
{
depthPass = depthPass || depthTest(r, zBuffer, q, x, z[q], sMask[q], zMask[q], cMask[q]);
depthPass = depthPass || depthTest(zBuffer, q, x, z[q], sMask[q], zMask[q], cMask[q]);
}
}
......@@ -258,11 +246,11 @@ namespace sw
{
if(state.multiSampleMask & (1 << q))
{
writeDepth(r, zBuffer, q, x, z[q], zMask[q]);
writeDepth(zBuffer, q, x, z[q], zMask[q]);
if(state.occlusionEnabled)
{
r.occlusion += *Pointer<UInt>(r.constants + OFFSET(Constants,occlusionCount) + 4 * (zMask[q] & sMask[q]));
occlusion += *Pointer<UInt>(constants + OFFSET(Constants,occlusionCount) + 4 * (zMask[q] & sMask[q]));
}
}
}
......@@ -273,12 +261,12 @@ namespace sw
AddAtomic(Pointer<Long>(&profiler.ropOperations), 4);
#endif
rasterOperation(r, f, cBuffer, x, sMask, zMask, cMask);
rasterOperation(f, cBuffer, x, sMask, zMask, cMask);
}
}
#if PERF_PROFILE
r.cycles[PERF_ROP] += Ticks() - ropTime;
cycles[PERF_ROP] += Ticks() - ropTime;
#endif
}
}
......@@ -287,12 +275,12 @@ namespace sw
{
if(state.multiSampleMask & (1 << q))
{
writeStencil(r, sBuffer, q, x, sMask[q], zMask[q], cMask[q]);
writeStencil(sBuffer, q, x, sMask[q], zMask[q], cMask[q]);
}
}
#if PERF_PROFILE
r.cycles[PERF_PIPE] += Ticks() - pipeTime;
cycles[PERF_PIPE] += Ticks() - pipeTime;
#endif
}
......@@ -314,7 +302,7 @@ namespace sw
return interpolant;
}
void PixelRoutine::stencilTest(Registers &r, Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &cMask)
void PixelRoutine::stencilTest(Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &cMask)
{
if(!state.stencilActive)
{
......@@ -327,7 +315,7 @@ namespace sw
if(q > 0)
{
buffer += q * *Pointer<Int>(r.data + OFFSET(DrawData,stencilSliceB));
buffer += q * *Pointer<Int>(data + OFFSET(DrawData,stencilSliceB));
}
Byte8 value = As<Byte8>(Long1(*Pointer<UInt>(buffer)));
......@@ -335,29 +323,29 @@ namespace sw
if(!state.noStencilMask)
{
value &= *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[0].testMaskQ));
value &= *Pointer<Byte8>(data + OFFSET(DrawData,stencil[0].testMaskQ));
}
stencilTest(r, value, state.stencilCompareMode, false);
stencilTest(value, state.stencilCompareMode, false);
if(state.twoSidedStencil)
{
if(!state.noStencilMaskCCW)
{
valueCCW &= *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[1].testMaskQ));
valueCCW &= *Pointer<Byte8>(data + OFFSET(DrawData,stencil[1].testMaskQ));
}
stencilTest(r, valueCCW, state.stencilCompareModeCCW, true);
stencilTest(valueCCW, state.stencilCompareModeCCW, true);
value &= *Pointer<Byte8>(r.primitive + OFFSET(Primitive,clockwiseMask));
valueCCW &= *Pointer<Byte8>(r.primitive + OFFSET(Primitive,invClockwiseMask));
value &= *Pointer<Byte8>(primitive + OFFSET(Primitive,clockwiseMask));
valueCCW &= *Pointer<Byte8>(primitive + OFFSET(Primitive,invClockwiseMask));
value |= valueCCW;
}
sMask = SignMask(value) & cMask;
}
void PixelRoutine::stencilTest(Registers &r, Byte8 &value, StencilCompareMode stencilCompareMode, bool CCW)
void PixelRoutine::stencilTest(Byte8 &value, StencilCompareMode stencilCompareMode, bool CCW)
{
Byte8 equal;
......@@ -371,31 +359,31 @@ namespace sw
break;
case STENCIL_LESS: // a < b ~ b > a
value += Byte8(0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80);
value = CmpGT(As<SByte8>(value), *Pointer<SByte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ)));
value = CmpGT(As<SByte8>(value), *Pointer<SByte8>(data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ)));
break;
case STENCIL_EQUAL:
value = CmpEQ(value, *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceMaskedQ)));
value = CmpEQ(value, *Pointer<Byte8>(data + OFFSET(DrawData,stencil[CCW].referenceMaskedQ)));
break;
case STENCIL_NOTEQUAL: // a != b ~ !(a == b)
value = CmpEQ(value, *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceMaskedQ)));
value = CmpEQ(value, *Pointer<Byte8>(data + OFFSET(DrawData,stencil[CCW].referenceMaskedQ)));
value ^= Byte8(0xFFFFFFFFFFFFFFFF);
break;
case STENCIL_LESSEQUAL: // a <= b ~ (b > a) || (a == b)
equal = value;
equal = CmpEQ(equal, *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceMaskedQ)));
equal = CmpEQ(equal, *Pointer<Byte8>(data + OFFSET(DrawData,stencil[CCW].referenceMaskedQ)));
value += Byte8(0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80);
value = CmpGT(As<SByte8>(value), *Pointer<SByte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ)));
value = CmpGT(As<SByte8>(value), *Pointer<SByte8>(data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ)));
value |= equal;
break;
case STENCIL_GREATER: // a > b
equal = *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ));
equal = *Pointer<Byte8>(data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ));
value += Byte8(0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80);
equal = CmpGT(As<SByte8>(equal), As<SByte8>(value));
value = equal;
break;
case STENCIL_GREATEREQUAL: // a >= b ~ !(a < b) ~ !(b > a)
value += Byte8(0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80);
value = CmpGT(As<SByte8>(value), *Pointer<SByte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ)));
value = CmpGT(As<SByte8>(value), *Pointer<SByte8>(data + OFFSET(DrawData,stencil[CCW].referenceMaskedSignedQ)));
value ^= Byte8(0xFFFFFFFFFFFFFFFF);
break;
default:
......@@ -403,7 +391,7 @@ namespace sw
}
}
Bool PixelRoutine::depthTest(Registers &r, Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &sMask, Int &zMask, Int &cMask)
Bool PixelRoutine::depthTest(Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &sMask, Int &zMask, Int &cMask)
{
if(!state.depthTestActive)
{
......@@ -416,11 +404,11 @@ namespace sw
{
if(complementaryDepthBuffer)
{
Z = Float4(1.0f) - r.oDepth;
Z = Float4(1.0f) - oDepth;
}
else
{
Z = r.oDepth;
Z = oDepth;
}
}
......@@ -430,7 +418,7 @@ namespace sw
if(!state.quadLayoutDepthBuffer)
{
buffer = zBuffer + 4 * x;
pitch = *Pointer<Int>(r.data + OFFSET(DrawData,depthPitchB));
pitch = *Pointer<Int>(data + OFFSET(DrawData,depthPitchB));
}
else
{
......@@ -439,7 +427,7 @@ namespace sw
if(q > 0)
{
buffer += q * *Pointer<Int>(r.data + OFFSET(DrawData,depthSliceB));
buffer += q * *Pointer<Int>(data + OFFSET(DrawData,depthSliceB));
}
Float4 zValue;
......@@ -539,7 +527,7 @@ namespace sw
return zMask != 0;
}
void PixelRoutine::alphaTest(Registers &r, Int &aMask, Short4 &alpha)
void PixelRoutine::alphaTest(Int &aMask, Short4 &alpha)
{
Short4 cmp;
Short4 equal;
......@@ -553,29 +541,29 @@ namespace sw
aMask = 0x0;
break;
case ALPHA_EQUAL:
cmp = CmpEQ(alpha, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.alphaReference4)));
cmp = CmpEQ(alpha, *Pointer<Short4>(data + OFFSET(DrawData,factor.alphaReference4)));
aMask = SignMask(Pack(cmp, Short4(0x0000, 0x0000, 0x0000, 0x0000)));
break;
case ALPHA_NOTEQUAL: // a != b ~ !(a == b)
cmp = CmpEQ(alpha, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.alphaReference4))) ^ Short4((short)0xFFFF, (short)0xFFFF, (short)0xFFFF, (short)0xFFFF); // FIXME
cmp = CmpEQ(alpha, *Pointer<Short4>(data + OFFSET(DrawData,factor.alphaReference4))) ^ Short4((short)0xFFFF, (short)0xFFFF, (short)0xFFFF, (short)0xFFFF); // FIXME
aMask = SignMask(Pack(cmp, Short4(0x0000, 0x0000, 0x0000, 0x0000)));
break;
case ALPHA_LESS: // a < b ~ b > a
cmp = CmpGT(*Pointer<Short4>(r.data + OFFSET(DrawData,factor.alphaReference4)), alpha);
cmp = CmpGT(*Pointer<Short4>(data + OFFSET(DrawData,factor.alphaReference4)), alpha);
aMask = SignMask(Pack(cmp, Short4(0x0000, 0x0000, 0x0000, 0x0000)));
break;
case ALPHA_GREATEREQUAL: // a >= b ~ (a > b) || (a == b) ~ !(b > a) // TODO: Approximate
equal = CmpEQ(alpha, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.alphaReference4)));
cmp = CmpGT(alpha, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.alphaReference4)));
equal = CmpEQ(alpha, *Pointer<Short4>(data + OFFSET(DrawData,factor.alphaReference4)));
cmp = CmpGT(alpha, *Pointer<Short4>(data + OFFSET(DrawData,factor.alphaReference4)));
cmp |= equal;
aMask = SignMask(Pack(cmp, Short4(0x0000, 0x0000, 0x0000, 0x0000)));
break;
case ALPHA_LESSEQUAL: // a <= b ~ !(a > b)
cmp = CmpGT(alpha, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.alphaReference4))) ^ Short4((short)0xFFFF, (short)0xFFFF, (short)0xFFFF, (short)0xFFFF); // FIXME
cmp = CmpGT(alpha, *Pointer<Short4>(data + OFFSET(DrawData,factor.alphaReference4))) ^ Short4((short)0xFFFF, (short)0xFFFF, (short)0xFFFF, (short)0xFFFF); // FIXME
aMask = SignMask(Pack(cmp, Short4(0x0000, 0x0000, 0x0000, 0x0000)));
break;
case ALPHA_GREATER: // a > b
cmp = CmpGT(alpha, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.alphaReference4)));
cmp = CmpGT(alpha, *Pointer<Short4>(data + OFFSET(DrawData,factor.alphaReference4)));
aMask = SignMask(Pack(cmp, Short4(0x0000, 0x0000, 0x0000, 0x0000)));
break;
default:
......@@ -583,12 +571,12 @@ namespace sw
}
}
void PixelRoutine::alphaToCoverage(Registers &r, Int cMask[4], Float4 &alpha)
void PixelRoutine::alphaToCoverage(Int cMask[4], Float4 &alpha)
{
Int4 coverage0 = CmpNLT(alpha, *Pointer<Float4>(r.data + OFFSET(DrawData,a2c0)));
Int4 coverage1 = CmpNLT(alpha, *Pointer<Float4>(r.data + OFFSET(DrawData,a2c1)));
Int4 coverage2 = CmpNLT(alpha, *Pointer<Float4>(r.data + OFFSET(DrawData,a2c2)));
Int4 coverage3 = CmpNLT(alpha, *Pointer<Float4>(r.data + OFFSET(DrawData,a2c3)));
Int4 coverage0 = CmpNLT(alpha, *Pointer<Float4>(data + OFFSET(DrawData,a2c0)));
Int4 coverage1 = CmpNLT(alpha, *Pointer<Float4>(data + OFFSET(DrawData,a2c1)));
Int4 coverage2 = CmpNLT(alpha, *Pointer<Float4>(data + OFFSET(DrawData,a2c2)));
Int4 coverage3 = CmpNLT(alpha, *Pointer<Float4>(data + OFFSET(DrawData,a2c3)));
Int aMask0 = SignMask(coverage0);
Int aMask1 = SignMask(coverage1);
......@@ -601,7 +589,7 @@ namespace sw
cMask[3] &= aMask3;
}
void PixelRoutine::fogBlend(Registers &r, Vector4f &c0, Float4 &fog)
void PixelRoutine::fogBlend(Vector4f &c0, Float4 &fog)
{
if(!state.fogActive)
{
......@@ -610,26 +598,26 @@ namespace sw
if(state.pixelFogMode != FOG_NONE)
{
pixelFog(r, fog);
pixelFog(fog);
fog = Min(fog, Float4(1.0f));
fog = Max(fog, Float4(0.0f));
}
c0.x -= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[0]));
c0.y -= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[1]));
c0.z -= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[2]));
c0.x -= *Pointer<Float4>(data + OFFSET(DrawData,fog.colorF[0]));
c0.y -= *Pointer<Float4>(data + OFFSET(DrawData,fog.colorF[1]));
c0.z -= *Pointer<Float4>(data + OFFSET(DrawData,fog.colorF[2]));
c0.x *= fog;
c0.y *= fog;
c0.z *= fog;
c0.x += *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[0]));
c0.y += *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[1]));
c0.z += *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[2]));
c0.x += *Pointer<Float4>(data + OFFSET(DrawData,fog.colorF[0]));
c0.y += *Pointer<Float4>(data + OFFSET(DrawData,fog.colorF[1]));
c0.z += *Pointer<Float4>(data + OFFSET(DrawData,fog.colorF[2]));
}
void PixelRoutine::pixelFog(Registers &r, Float4 &visibility)
void PixelRoutine::pixelFog(Float4 &visibility)
{
Float4 &zw = visibility;
......@@ -637,17 +625,17 @@ namespace sw
{
if(state.wBasedFog)
{
zw = r.rhw;
zw = rhw;
}
else
{
if(complementaryDepthBuffer)
{
zw = Float4(1.0f) - r.z[0];
zw = Float4(1.0f) - z[0];
}
else
{
zw = r.z[0];
zw = z[0];
}
}
}
......@@ -657,16 +645,16 @@ namespace sw
case FOG_NONE:
break;
case FOG_LINEAR:
zw *= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.scale));
zw += *Pointer<Float4>(r.data + OFFSET(DrawData,fog.offset));
zw *= *Pointer<Float4>(data + OFFSET(DrawData,fog.scale));
zw += *Pointer<Float4>(data + OFFSET(DrawData,fog.offset));
break;
case FOG_EXP:
zw *= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.densityE));
zw *= *Pointer<Float4>(data + OFFSET(DrawData,fog.densityE));
zw = exponential2(zw, true);
break;
case FOG_EXP2:
zw *= zw;
zw *= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.density2E));
zw *= *Pointer<Float4>(data + OFFSET(DrawData,fog.density2E));
zw = exponential2(zw, true);
break;
default:
......@@ -674,7 +662,7 @@ namespace sw
}
}
void PixelRoutine::writeDepth(Registers &r, Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &zMask)
void PixelRoutine::writeDepth(Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &zMask)
{
if(!state.depthWriteEnable)
{
......@@ -687,11 +675,11 @@ namespace sw
{
if(complementaryDepthBuffer)
{
Z = Float4(1.0f) - r.oDepth;
Z = Float4(1.0f) - oDepth;
}
else
{
Z = r.oDepth;
Z = oDepth;
}
}
......@@ -701,7 +689,7 @@ namespace sw
if(!state.quadLayoutDepthBuffer)
{
buffer = zBuffer + 4 * x;
pitch = *Pointer<Int>(r.data + OFFSET(DrawData,depthPitchB));
pitch = *Pointer<Int>(data + OFFSET(DrawData,depthPitchB));
}
else
{
......@@ -710,7 +698,7 @@ namespace sw
if(q > 0)
{
buffer += q * *Pointer<Int>(r.data + OFFSET(DrawData,depthSliceB));
buffer += q * *Pointer<Int>(data + OFFSET(DrawData,depthSliceB));
}
Float4 zValue;
......@@ -729,8 +717,8 @@ namespace sw
}
}
Z = As<Float4>(As<Int4>(Z) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X) + zMask * 16, 16));
zValue = As<Float4>(As<Int4>(zValue) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD4X) + zMask * 16, 16));
Z = As<Float4>(As<Int4>(Z) & *Pointer<Int4>(constants + OFFSET(Constants,maskD4X) + zMask * 16, 16));
zValue = As<Float4>(As<Int4>(zValue) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD4X) + zMask * 16, 16));
Z = As<Float4>(As<Int4>(Z) | As<Int4>(zValue));
if(!state.quadLayoutDepthBuffer)
......@@ -745,7 +733,7 @@ namespace sw
}
}
void PixelRoutine::writeStencil(Registers &r, Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &zMask, Int &cMask)
void PixelRoutine::writeStencil(Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &zMask, Int &cMask)
{
if(!state.stencilActive)
{
......@@ -769,19 +757,19 @@ namespace sw
if(q > 0)
{
buffer += q * *Pointer<Int>(r.data + OFFSET(DrawData,stencilSliceB));
buffer += q * *Pointer<Int>(data + OFFSET(DrawData,stencilSliceB));
}
Byte8 bufferValue = As<Byte8>(Long1(*Pointer<UInt>(buffer)));
Byte8 newValue;
stencilOperation(r, newValue, bufferValue, state.stencilPassOperation, state.stencilZFailOperation, state.stencilFailOperation, false, zMask, sMask);
stencilOperation(newValue, bufferValue, state.stencilPassOperation, state.stencilZFailOperation, state.stencilFailOperation, false, zMask, sMask);
if(!state.noStencilWriteMask)
{
Byte8 maskedValue = bufferValue;
newValue &= *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[0].writeMaskQ));
maskedValue &= *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[0].invWriteMaskQ));
newValue &= *Pointer<Byte8>(data + OFFSET(DrawData,stencil[0].writeMaskQ));
maskedValue &= *Pointer<Byte8>(data + OFFSET(DrawData,stencil[0].invWriteMaskQ));
newValue |= maskedValue;
}
......@@ -789,62 +777,62 @@ namespace sw
{
Byte8 newValueCCW;
stencilOperation(r, newValueCCW, bufferValue, state.stencilPassOperationCCW, state.stencilZFailOperationCCW, state.stencilFailOperationCCW, true, zMask, sMask);
stencilOperation(newValueCCW, bufferValue, state.stencilPassOperationCCW, state.stencilZFailOperationCCW, state.stencilFailOperationCCW, true, zMask, sMask);
if(!state.noStencilWriteMaskCCW)
{
Byte8 maskedValue = bufferValue;
newValueCCW &= *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[1].writeMaskQ));
maskedValue &= *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[1].invWriteMaskQ));
newValueCCW &= *Pointer<Byte8>(data + OFFSET(DrawData,stencil[1].writeMaskQ));
maskedValue &= *Pointer<Byte8>(data + OFFSET(DrawData,stencil[1].invWriteMaskQ));
newValueCCW |= maskedValue;
}
newValue &= *Pointer<Byte8>(r.primitive + OFFSET(Primitive,clockwiseMask));
newValueCCW &= *Pointer<Byte8>(r.primitive + OFFSET(Primitive,invClockwiseMask));
newValue &= *Pointer<Byte8>(primitive + OFFSET(Primitive,clockwiseMask));
newValueCCW &= *Pointer<Byte8>(primitive + OFFSET(Primitive,invClockwiseMask));
newValue |= newValueCCW;
}
newValue &= *Pointer<Byte8>(r.constants + OFFSET(Constants,maskB4Q) + 8 * cMask);
bufferValue &= *Pointer<Byte8>(r.constants + OFFSET(Constants,invMaskB4Q) + 8 * cMask);
newValue &= *Pointer<Byte8>(constants + OFFSET(Constants,maskB4Q) + 8 * cMask);
bufferValue &= *Pointer<Byte8>(constants + OFFSET(Constants,invMaskB4Q) + 8 * cMask);
newValue |= bufferValue;
*Pointer<UInt>(buffer) = UInt(As<Long>(newValue));
}
void PixelRoutine::stencilOperation(Registers &r, Byte8 &newValue, Byte8 &bufferValue, StencilOperation stencilPassOperation, StencilOperation stencilZFailOperation, StencilOperation stencilFailOperation, bool CCW, Int &zMask, Int &sMask)
void PixelRoutine::stencilOperation(Byte8 &newValue, Byte8 &bufferValue, StencilOperation stencilPassOperation, StencilOperation stencilZFailOperation, StencilOperation stencilFailOperation, bool CCW, Int &zMask, Int &sMask)
{
Byte8 &pass = newValue;
Byte8 fail;
Byte8 zFail;
stencilOperation(r, pass, bufferValue, stencilPassOperation, CCW);
stencilOperation(pass, bufferValue, stencilPassOperation, CCW);
if(stencilZFailOperation != stencilPassOperation)
{
stencilOperation(r, zFail, bufferValue, stencilZFailOperation, CCW);
stencilOperation(zFail, bufferValue, stencilZFailOperation, CCW);
}
if(stencilFailOperation != stencilPassOperation || stencilFailOperation != stencilZFailOperation)
{
stencilOperation(r, fail, bufferValue, stencilFailOperation, CCW);
stencilOperation(fail, bufferValue, stencilFailOperation, CCW);
}
if(stencilFailOperation != stencilPassOperation || stencilFailOperation != stencilZFailOperation)
{
if(state.depthTestActive && stencilZFailOperation != stencilPassOperation) // zMask valid and values not the same
{
pass &= *Pointer<Byte8>(r.constants + OFFSET(Constants,maskB4Q) + 8 * zMask);
zFail &= *Pointer<Byte8>(r.constants + OFFSET(Constants,invMaskB4Q) + 8 * zMask);
pass &= *Pointer<Byte8>(constants + OFFSET(Constants,maskB4Q) + 8 * zMask);
zFail &= *Pointer<Byte8>(constants + OFFSET(Constants,invMaskB4Q) + 8 * zMask);
pass |= zFail;
}
pass &= *Pointer<Byte8>(r.constants + OFFSET(Constants,maskB4Q) + 8 * sMask);
fail &= *Pointer<Byte8>(r.constants + OFFSET(Constants,invMaskB4Q) + 8 * sMask);
pass &= *Pointer<Byte8>(constants + OFFSET(Constants,maskB4Q) + 8 * sMask);
fail &= *Pointer<Byte8>(constants + OFFSET(Constants,invMaskB4Q) + 8 * sMask);
pass |= fail;
}
}
void PixelRoutine::stencilOperation(Registers &r, Byte8 &output, Byte8 &bufferValue, StencilOperation operation, bool CCW)
void PixelRoutine::stencilOperation(Byte8 &output, Byte8 &bufferValue, StencilOperation operation, bool CCW)
{
switch(operation)
{
......@@ -855,7 +843,7 @@ namespace sw
output = Byte8(0x0000000000000000);
break;
case OPERATION_REPLACE:
output = *Pointer<Byte8>(r.data + OFFSET(DrawData,stencil[CCW].referenceQ));
output = *Pointer<Byte8>(data + OFFSET(DrawData,stencil[CCW].referenceQ));
break;
case OPERATION_INCRSAT:
output = AddSat(bufferValue, Byte8(1, 1, 1, 1, 1, 1, 1, 1));
......@@ -877,7 +865,7 @@ namespace sw
}
}
void PixelRoutine::blendFactor(Registers &r, const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorActive)
void PixelRoutine::blendFactor(const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorActive)
{
switch(blendFactorActive)
{
......@@ -934,31 +922,31 @@ namespace sw
blendFactor.z = blendFactor.x;
break;
case BLEND_CONSTANT:
blendFactor.x = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[0]));
blendFactor.y = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[1]));
blendFactor.z = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[2]));
blendFactor.x = *Pointer<Short4>(data + OFFSET(DrawData,factor.blendConstant4W[0]));
blendFactor.y = *Pointer<Short4>(data + OFFSET(DrawData,factor.blendConstant4W[1]));
blendFactor.z = *Pointer<Short4>(data + OFFSET(DrawData,factor.blendConstant4W[2]));
break;
case BLEND_INVCONSTANT:
blendFactor.x = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[0]));
blendFactor.y = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[1]));
blendFactor.z = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[2]));
blendFactor.x = *Pointer<Short4>(data + OFFSET(DrawData,factor.invBlendConstant4W[0]));
blendFactor.y = *Pointer<Short4>(data + OFFSET(DrawData,factor.invBlendConstant4W[1]));
blendFactor.z = *Pointer<Short4>(data + OFFSET(DrawData,factor.invBlendConstant4W[2]));
break;
case BLEND_CONSTANTALPHA:
blendFactor.x = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
blendFactor.y = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
blendFactor.z = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
blendFactor.x = *Pointer<Short4>(data + OFFSET(DrawData,factor.blendConstant4W[3]));
blendFactor.y = *Pointer<Short4>(data + OFFSET(DrawData,factor.blendConstant4W[3]));
blendFactor.z = *Pointer<Short4>(data + OFFSET(DrawData,factor.blendConstant4W[3]));
break;
case BLEND_INVCONSTANTALPHA:
blendFactor.x = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
blendFactor.y = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
blendFactor.z = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
blendFactor.x = *Pointer<Short4>(data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
blendFactor.y = *Pointer<Short4>(data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
blendFactor.z = *Pointer<Short4>(data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
break;
default:
ASSERT(false);
}
}
void PixelRoutine::blendFactorAlpha(Registers &r, const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorAlphaActive)
void PixelRoutine::blendFactorAlpha(const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorAlphaActive)
{
switch(blendFactorAlphaActive)
{
......@@ -997,18 +985,18 @@ namespace sw
break;
case BLEND_CONSTANT:
case BLEND_CONSTANTALPHA:
blendFactor.w = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
blendFactor.w = *Pointer<Short4>(data + OFFSET(DrawData,factor.blendConstant4W[3]));
break;
case BLEND_INVCONSTANT:
case BLEND_INVCONSTANTALPHA:
blendFactor.w = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
blendFactor.w = *Pointer<Short4>(data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
break;
default:
ASSERT(false);
}
}
void PixelRoutine::readPixel(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Vector4s &pixel)
void PixelRoutine::readPixel(int index, Pointer<Byte> &cBuffer, Int &x, Vector4s &pixel)
{
Short4 c01;
Short4 c23;
......@@ -1019,7 +1007,7 @@ namespace sw
{
case FORMAT_R5G6B5:
buffer = cBuffer + 2 * x;
buffer2 = buffer + *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer2 = buffer + *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
c01 = As<Short4>(Int2(*Pointer<Int>(buffer), *Pointer<Int>(buffer2)));
pixel.x = c01 & Short4(0xF800u);
......@@ -1030,7 +1018,7 @@ namespace sw
case FORMAT_A8R8G8B8:
buffer = cBuffer + 4 * x;
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
pixel.z = c01;
pixel.y = c01;
......@@ -1049,7 +1037,7 @@ namespace sw
case FORMAT_A8B8G8R8:
buffer = cBuffer + 4 * x;
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
pixel.z = c01;
pixel.y = c01;
......@@ -1068,7 +1056,7 @@ namespace sw
case FORMAT_A8:
buffer = cBuffer + 1 * x;
pixel.w = Insert(pixel.w, *Pointer<Short>(buffer), 0);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
pixel.w = Insert(pixel.w, *Pointer<Short>(buffer), 1);
pixel.w = UnpackLow(As<Byte8>(pixel.w), As<Byte8>(pixel.w));
pixel.x = Short4(0x0000);
......@@ -1078,7 +1066,7 @@ namespace sw
case FORMAT_X8R8G8B8:
buffer = cBuffer + 4 * x;
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
pixel.z = c01;
pixel.y = c01;
......@@ -1096,7 +1084,7 @@ namespace sw
case FORMAT_X8B8G8R8:
buffer = cBuffer + 4 * x;
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
pixel.z = c01;
pixel.y = c01;
......@@ -1130,7 +1118,7 @@ namespace sw
buffer = cBuffer;
pixel.x = *Pointer<Short4>(buffer + 8 * x);
pixel.y = *Pointer<Short4>(buffer + 8 * x + 8);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
pixel.z = *Pointer<Short4>(buffer + 8 * x);
pixel.w = *Pointer<Short4>(buffer + 8 * x + 8);
transpose4x4(pixel.x, pixel.y, pixel.z, pixel.w);
......@@ -1138,7 +1126,7 @@ namespace sw
case FORMAT_G16R16:
buffer = cBuffer;
pixel.x = *Pointer<Short4>(buffer + 4 * x);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData, colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData, colorPitchB[index]));
pixel.y = *Pointer<Short4>(buffer + 4 * x);
pixel.z = pixel.x;
pixel.x = As<Short4>(UnpackLow(pixel.x, pixel.y));
......@@ -1155,11 +1143,11 @@ namespace sw
if(postBlendSRGB && state.writeSRGB)
{
sRGBtoLinear16_12_16(r, pixel);
sRGBtoLinear16_12_16(pixel);
}
}
void PixelRoutine::alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x)
void PixelRoutine::alphaBlend(int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x)
{
if(!state.alphaBlendActive)
{
......@@ -1167,14 +1155,14 @@ namespace sw
}
Vector4s pixel;
readPixel(r, index, cBuffer, x, pixel);
readPixel(index, cBuffer, x, pixel);
// Final Color = ObjectColor * SourceBlendFactor + PixelColor * DestinationBlendFactor
Vector4s sourceFactor;
Vector4s destFactor;
blendFactor(r, sourceFactor, current, pixel, state.sourceBlendFactor);
blendFactor(r, destFactor, current, pixel, state.destBlendFactor);
blendFactor(sourceFactor, current, pixel, state.sourceBlendFactor);
blendFactor(destFactor, current, pixel, state.destBlendFactor);
if(state.sourceBlendFactor != BLEND_ONE && state.sourceBlendFactor != BLEND_ZERO)
{
......@@ -1234,8 +1222,8 @@ namespace sw
ASSERT(false);
}
blendFactorAlpha(r, sourceFactor, current, pixel, state.sourceBlendFactorAlpha);
blendFactorAlpha(r, destFactor, current, pixel, state.destBlendFactorAlpha);
blendFactorAlpha(sourceFactor, current, pixel, state.sourceBlendFactorAlpha);
blendFactorAlpha(destFactor, current, pixel, state.destBlendFactorAlpha);
if(state.sourceBlendFactorAlpha != BLEND_ONE && state.sourceBlendFactorAlpha != BLEND_ZERO)
{
......@@ -1278,7 +1266,7 @@ namespace sw
}
}
void PixelRoutine::logicOperation(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x)
void PixelRoutine::logicOperation(int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x)
{
if(state.logicalOperation == LOGICALOP_COPY)
{
......@@ -1286,7 +1274,7 @@ namespace sw
}
Vector4s pixel;
readPixel(r, index, cBuffer, x, pixel);
readPixel(index, cBuffer, x, pixel);
switch(state.logicalOperation)
{
......@@ -1373,11 +1361,11 @@ namespace sw
}
}
void PixelRoutine::writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Vector4s &current, Int &sMask, Int &zMask, Int &cMask)
void PixelRoutine::writeColor(int index, Pointer<Byte> &cBuffer, Int &x, Vector4s &current, Int &sMask, Int &zMask, Int &cMask)
{
if(postBlendSRGB && state.writeSRGB)
{
linearToSRGB16_12_16(r, current);
linearToSRGB16_12_16(current);
}
if(exactColorRounding)
......@@ -1559,17 +1547,17 @@ namespace sw
if((bgraWriteMask & 0x00000007) != 0x00000007)
{
Int masked = value;
c01 &= *Pointer<Int>(r.constants + OFFSET(Constants,mask565Q[bgraWriteMask & 0x7][0]));
masked &= *Pointer<Int>(r.constants + OFFSET(Constants,invMask565Q[bgraWriteMask & 0x7][0]));
c01 &= *Pointer<Int>(constants + OFFSET(Constants,mask565Q[bgraWriteMask & 0x7][0]));
masked &= *Pointer<Int>(constants + OFFSET(Constants,invMask565Q[bgraWriteMask & 0x7][0]));
c01 |= masked;
}
c01 &= *Pointer<Int>(r.constants + OFFSET(Constants,maskW4Q[0][0]) + xMask * 8);
value &= *Pointer<Int>(r.constants + OFFSET(Constants,invMaskW4Q[0][0]) + xMask * 8);
c01 &= *Pointer<Int>(constants + OFFSET(Constants,maskW4Q[0][0]) + xMask * 8);
value &= *Pointer<Int>(constants + OFFSET(Constants,invMaskW4Q[0][0]) + xMask * 8);
c01 |= value;
*Pointer<Int>(buffer) = c01;
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
value = *Pointer<Int>(buffer);
Int c23 = Extract(As<Int2>(current.x), 1);
......@@ -1577,13 +1565,13 @@ namespace sw
if((bgraWriteMask & 0x00000007) != 0x00000007)
{
Int masked = value;
c23 &= *Pointer<Int>(r.constants + OFFSET(Constants,mask565Q[bgraWriteMask & 0x7][0]));
masked &= *Pointer<Int>(r.constants + OFFSET(Constants,invMask565Q[bgraWriteMask & 0x7][0]));
c23 &= *Pointer<Int>(constants + OFFSET(Constants,mask565Q[bgraWriteMask & 0x7][0]));
masked &= *Pointer<Int>(constants + OFFSET(Constants,invMask565Q[bgraWriteMask & 0x7][0]));
c23 |= masked;
}
c23 &= *Pointer<Int>(r.constants + OFFSET(Constants,maskW4Q[0][2]) + xMask * 8);
value &= *Pointer<Int>(r.constants + OFFSET(Constants,invMaskW4Q[0][2]) + xMask * 8);
c23 &= *Pointer<Int>(constants + OFFSET(Constants,maskW4Q[0][2]) + xMask * 8);
value &= *Pointer<Int>(constants + OFFSET(Constants,invMaskW4Q[0][2]) + xMask * 8);
c23 |= value;
*Pointer<Int>(buffer) = c23;
}
......@@ -1598,13 +1586,13 @@ namespace sw
// (state.targetFormat[index] == FORMAT_X8G8R8B8Q && bgraWriteMask != 0x0000000F))) // FIXME: Need for masking when XRGB && Fh?
// {
// Short4 masked = value;
// c01 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
// masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
// c01 &= *Pointer<Short4>(constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
// masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
// c01 |= masked;
// }
// c01 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD01Q) + xMask * 8);
// value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
// c01 &= *Pointer<Short4>(constants + OFFSET(Constants,maskD01Q) + xMask * 8);
// value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
// c01 |= value;
// *Pointer<Short4>(cBuffer + 8 * x + 0) = c01;
......@@ -1615,13 +1603,13 @@ namespace sw
// (state.targetFormat[index] == FORMAT_X8G8R8B8Q && bgraWriteMask != 0x0000000F))) // FIXME: Need for masking when XRGB && Fh?
// {
// Short4 masked = value;
// c23 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
// masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
// c23 &= *Pointer<Short4>(constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
// masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
// c23 |= masked;
// }
// c23 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD23Q) + xMask * 8);
// value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
// c23 &= *Pointer<Short4>(constants + OFFSET(Constants,maskD23Q) + xMask * 8);
// value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
// c23 |= value;
// *Pointer<Short4>(cBuffer + 8 * x + 8) = c23;
break;
......@@ -1636,17 +1624,17 @@ namespace sw
(state.targetFormat[index] == FORMAT_X8R8G8B8 && bgraWriteMask != 0x0000000F))) // FIXME: Need for masking when XRGB && Fh?
{
Short4 masked = value;
c01 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
c01 &= *Pointer<Short4>(constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
c01 |= masked;
}
c01 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD01Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
c01 &= *Pointer<Short4>(constants + OFFSET(Constants,maskD01Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
c01 |= value;
*Pointer<Short4>(buffer) = c01;
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
value = *Pointer<Short4>(buffer);
if((state.targetFormat[index] == FORMAT_A8R8G8B8 && bgraWriteMask != 0x0000000F) ||
......@@ -1654,13 +1642,13 @@ namespace sw
(state.targetFormat[index] == FORMAT_X8R8G8B8 && bgraWriteMask != 0x0000000F))) // FIXME: Need for masking when XRGB && Fh?
{
Short4 masked = value;
c23 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
c23 &= *Pointer<Short4>(constants + OFFSET(Constants,maskB4Q[bgraWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskB4Q[bgraWriteMask][0]));
c23 |= masked;
}
c23 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD23Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
c23 &= *Pointer<Short4>(constants + OFFSET(Constants,maskD23Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
c23 |= value;
*Pointer<Short4>(buffer) = c23;
}
......@@ -1676,17 +1664,17 @@ namespace sw
(state.targetFormat[index] == FORMAT_X8B8G8R8 && rgbaWriteMask != 0x0000000F))) // FIXME: Need for masking when XBGR && Fh?
{
Short4 masked = value;
c01 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskB4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskB4Q[rgbaWriteMask][0]));
c01 &= *Pointer<Short4>(constants + OFFSET(Constants,maskB4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskB4Q[rgbaWriteMask][0]));
c01 |= masked;
}
c01 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD01Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
c01 &= *Pointer<Short4>(constants + OFFSET(Constants,maskD01Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
c01 |= value;
*Pointer<Short4>(buffer) = c01;
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
value = *Pointer<Short4>(buffer);
if((state.targetFormat[index] == FORMAT_A8B8G8R8 && rgbaWriteMask != 0x0000000F) ||
......@@ -1694,13 +1682,13 @@ namespace sw
(state.targetFormat[index] == FORMAT_X8B8G8R8 && rgbaWriteMask != 0x0000000F))) // FIXME: Need for masking when XBGR && Fh?
{
Short4 masked = value;
c23 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskB4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskB4Q[rgbaWriteMask][0]));
c23 &= *Pointer<Short4>(constants + OFFSET(Constants,maskB4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskB4Q[rgbaWriteMask][0]));
c23 |= masked;
}
c23 &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD23Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
c23 &= *Pointer<Short4>(constants + OFFSET(Constants,maskD23Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
c23 |= value;
*Pointer<Short4>(buffer) = c23;
}
......@@ -1711,12 +1699,12 @@ namespace sw
Pointer<Byte> buffer = cBuffer + 1 * x;
Short4 value;
Insert(value, *Pointer<Short>(buffer), 0);
Int pitch = *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
Int pitch = *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
Insert(value, *Pointer<Short>(buffer + pitch), 1);
value = UnpackLow(As<Byte8>(value), As<Byte8>(value));
current.w &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskB4Q) + 8 * xMask);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskB4Q) + 8 * xMask);
current.w &= *Pointer<Short4>(constants + OFFSET(Constants,maskB4Q) + 8 * xMask);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskB4Q) + 8 * xMask);
current.w |= value;
*Pointer<Short>(buffer) = Extract(current.w, 0);
......@@ -1732,30 +1720,30 @@ namespace sw
if((rgbaWriteMask & 0x00000003) != 0x00000003)
{
Short4 masked = value;
current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW01Q[rgbaWriteMask & 0x3][0]));
current.x &= *Pointer<Short4>(constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskW01Q[rgbaWriteMask & 0x3][0]));
current.x |= masked;
}
current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD01Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
current.x &= *Pointer<Short4>(constants + OFFSET(Constants,maskD01Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
current.x |= value;
*Pointer<Short4>(buffer) = current.x;
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
value = *Pointer<Short4>(buffer);
if((rgbaWriteMask & 0x00000003) != 0x00000003)
{
Short4 masked = value;
current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW01Q[rgbaWriteMask & 0x3][0]));
current.y &= *Pointer<Short4>(constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskW01Q[rgbaWriteMask & 0x3][0]));
current.y |= masked;
}
current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD23Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
current.y &= *Pointer<Short4>(constants + OFFSET(Constants,maskD23Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
current.y |= value;
*Pointer<Short4>(buffer) = current.y;
}
......@@ -1770,13 +1758,13 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.x &= *Pointer<Short4>(constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.x |= masked;
}
current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ0Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ0Q) + xMask * 8);
current.x &= *Pointer<Short4>(constants + OFFSET(Constants,maskQ0Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskQ0Q) + xMask * 8);
current.x |= value;
*Pointer<Short4>(buffer) = current.x;
}
......@@ -1787,18 +1775,18 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.y &= *Pointer<Short4>(constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.y |= masked;
}
current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ1Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ1Q) + xMask * 8);
current.y &= *Pointer<Short4>(constants + OFFSET(Constants,maskQ1Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskQ1Q) + xMask * 8);
current.y |= value;
*Pointer<Short4>(buffer + 8) = current.y;
}
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
{
Short4 value = *Pointer<Short4>(buffer);
......@@ -1806,13 +1794,13 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
current.z &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.z &= *Pointer<Short4>(constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.z |= masked;
}
current.z &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ2Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ2Q) + xMask * 8);
current.z &= *Pointer<Short4>(constants + OFFSET(Constants,maskQ2Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskQ2Q) + xMask * 8);
current.z |= value;
*Pointer<Short4>(buffer) = current.z;
}
......@@ -1823,13 +1811,13 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
current.w &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.w &= *Pointer<Short4>(constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
current.w |= masked;
}
current.w &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ3Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ3Q) + xMask * 8);
current.w &= *Pointer<Short4>(constants + OFFSET(Constants,maskQ3Q) + xMask * 8);
value &= *Pointer<Short4>(constants + OFFSET(Constants,invMaskQ3Q) + xMask * 8);
current.w |= value;
*Pointer<Short4>(buffer + 8) = current.w;
}
......@@ -1840,7 +1828,7 @@ namespace sw
}
}
void PixelRoutine::blendFactor(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorActive)
void PixelRoutine::blendFactor(const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorActive)
{
switch(blendFactorActive)
{
......@@ -1897,21 +1885,21 @@ namespace sw
blendFactor.z = blendFactor.x;
break;
case BLEND_CONSTANT:
blendFactor.x = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[0]));
blendFactor.y = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[1]));
blendFactor.z = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[2]));
blendFactor.x = *Pointer<Float4>(data + OFFSET(DrawData,factor.blendConstant4F[0]));
blendFactor.y = *Pointer<Float4>(data + OFFSET(DrawData,factor.blendConstant4F[1]));
blendFactor.z = *Pointer<Float4>(data + OFFSET(DrawData,factor.blendConstant4F[2]));
break;
case BLEND_INVCONSTANT:
blendFactor.x = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[0]));
blendFactor.y = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[1]));
blendFactor.z = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[2]));
blendFactor.x = *Pointer<Float4>(data + OFFSET(DrawData,factor.invBlendConstant4F[0]));
blendFactor.y = *Pointer<Float4>(data + OFFSET(DrawData,factor.invBlendConstant4F[1]));
blendFactor.z = *Pointer<Float4>(data + OFFSET(DrawData,factor.invBlendConstant4F[2]));
break;
default:
ASSERT(false);
}
}
void PixelRoutine::blendFactorAlpha(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorAlphaActive)
void PixelRoutine::blendFactorAlpha(const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorAlphaActive)
{
switch(blendFactorAlphaActive)
{
......@@ -1949,17 +1937,17 @@ namespace sw
blendFactor.w = Float4(1.0f);
break;
case BLEND_CONSTANT:
blendFactor.w = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[3]));
blendFactor.w = *Pointer<Float4>(data + OFFSET(DrawData,factor.blendConstant4F[3]));
break;
case BLEND_INVCONSTANT:
blendFactor.w = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[3]));
blendFactor.w = *Pointer<Float4>(data + OFFSET(DrawData,factor.invBlendConstant4F[3]));
break;
default:
ASSERT(false);
}
}
void PixelRoutine::alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4f &oC, Int &x)
void PixelRoutine::alphaBlend(int index, Pointer<Byte> &cBuffer, Vector4f &oC, Int &x)
{
if(!state.alphaBlendActive)
{
......@@ -1980,7 +1968,7 @@ namespace sw
// FIXME: movlps
pixel.x.x = *Pointer<Float>(buffer + 4 * x + 0);
pixel.x.y = *Pointer<Float>(buffer + 4 * x + 4);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
// FIXME: movhps
pixel.x.z = *Pointer<Float>(buffer + 4 * x + 0);
pixel.x.w = *Pointer<Float>(buffer + 4 * x + 4);
......@@ -1991,7 +1979,7 @@ namespace sw
case FORMAT_G32R32F:
buffer = cBuffer;
pixel.x = *Pointer<Float4>(buffer + 8 * x, 16);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
pixel.y = *Pointer<Float4>(buffer + 8 * x, 16);
pixel.z = pixel.x;
pixel.x = ShuffleLowHigh(pixel.x, pixel.y, 0x88);
......@@ -2004,7 +1992,7 @@ namespace sw
buffer = cBuffer;
pixel.x = *Pointer<Float4>(buffer + 16 * x, 16);
pixel.y = *Pointer<Float4>(buffer + 16 * x + 16, 16);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
pixel.z = *Pointer<Float4>(buffer + 16 * x, 16);
pixel.w = *Pointer<Float4>(buffer + 16 * x + 16, 16);
transpose4x4(pixel.x, pixel.y, pixel.z, pixel.w);
......@@ -2024,8 +2012,8 @@ namespace sw
Vector4f sourceFactor;
Vector4f destFactor;
blendFactor(r, sourceFactor, oC, pixel, state.sourceBlendFactor);
blendFactor(r, destFactor, oC, pixel, state.destBlendFactor);
blendFactor(sourceFactor, oC, pixel, state.sourceBlendFactor);
blendFactor(destFactor, oC, pixel, state.destBlendFactor);
if(state.sourceBlendFactor != BLEND_ONE && state.sourceBlendFactor != BLEND_ZERO)
{
......@@ -2085,8 +2073,8 @@ namespace sw
ASSERT(false);
}
blendFactorAlpha(r, sourceFactor, oC, pixel, state.sourceBlendFactorAlpha);
blendFactorAlpha(r, destFactor, oC, pixel, state.destBlendFactorAlpha);
blendFactorAlpha(sourceFactor, oC, pixel, state.sourceBlendFactorAlpha);
blendFactorAlpha(destFactor, oC, pixel, state.destBlendFactorAlpha);
if(state.sourceBlendFactorAlpha != BLEND_ONE && state.sourceBlendFactorAlpha != BLEND_ZERO)
{
......@@ -2130,7 +2118,7 @@ namespace sw
}
}
void PixelRoutine::writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Vector4f &oC, Int &sMask, Int &zMask, Int &cMask)
void PixelRoutine::writeColor(int index, Pointer<Byte> &cBuffer, Int &x, Vector4f &oC, Int &sMask, Int &zMask, Int &cMask)
{
switch(state.targetFormat[index])
{
......@@ -2181,21 +2169,21 @@ namespace sw
value.x = *Pointer<Float>(buffer + 0);
value.y = *Pointer<Float>(buffer + 4);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
// FIXME: movhps
value.z = *Pointer<Float>(buffer + 0);
value.w = *Pointer<Float>(buffer + 4);
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD4X) + xMask * 16, 16));
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(constants + OFFSET(Constants,maskD4X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD4X) + xMask * 16, 16));
oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(value));
// FIXME: movhps
*Pointer<Float>(buffer + 0) = oC.x.z;
*Pointer<Float>(buffer + 4) = oC.x.w;
buffer -= *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer -= *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
// FIXME: movlps
*Pointer<Float>(buffer + 0) = oC.x.x;
......@@ -2210,17 +2198,17 @@ namespace sw
if((rgbaWriteMask & 0x00000003) != 0x00000003)
{
Float4 masked = value;
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD01X[rgbaWriteMask & 0x3][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD01X[rgbaWriteMask & 0x3][0])));
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(constants + OFFSET(Constants,maskD01X[rgbaWriteMask & 0x3][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD01X[rgbaWriteMask & 0x3][0])));
oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(masked));
}
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskQ01X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskQ01X) + xMask * 16, 16));
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(constants + OFFSET(Constants,maskQ01X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskQ01X) + xMask * 16, 16));
oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(value));
*Pointer<Float4>(buffer) = oC.x;
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
value = *Pointer<Float4>(buffer);
......@@ -2229,13 +2217,13 @@ namespace sw
Float4 masked;
masked = value;
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD01X[rgbaWriteMask & 0x3][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD01X[rgbaWriteMask & 0x3][0])));
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(constants + OFFSET(Constants,maskD01X[rgbaWriteMask & 0x3][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD01X[rgbaWriteMask & 0x3][0])));
oC.y = As<Float4>(As<Int4>(oC.y) | As<Int4>(masked));
}
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskQ23X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskQ23X) + xMask * 16, 16));
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(constants + OFFSET(Constants,maskQ23X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskQ23X) + xMask * 16, 16));
oC.y = As<Float4>(As<Int4>(oC.y) | As<Int4>(value));
*Pointer<Float4>(buffer) = oC.y;
break;
......@@ -2248,13 +2236,13 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(masked));
}
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX0X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskX0X) + xMask * 16, 16));
oC.x = As<Float4>(As<Int4>(oC.x) & *Pointer<Int4>(constants + OFFSET(Constants,maskX0X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskX0X) + xMask * 16, 16));
oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(value));
*Pointer<Float4>(buffer, 16) = oC.x;
}
......@@ -2265,18 +2253,18 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.y = As<Float4>(As<Int4>(oC.y) | As<Int4>(masked));
}
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX1X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskX1X) + xMask * 16, 16));
oC.y = As<Float4>(As<Int4>(oC.y) & *Pointer<Int4>(constants + OFFSET(Constants,maskX1X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskX1X) + xMask * 16, 16));
oC.y = As<Float4>(As<Int4>(oC.y) | As<Int4>(value));
*Pointer<Float4>(buffer + 16, 16) = oC.y;
}
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
buffer += *Pointer<Int>(data + OFFSET(DrawData,colorPitchB[index]));
{
value = *Pointer<Float4>(buffer, 16);
......@@ -2284,13 +2272,13 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
oC.z = As<Float4>(As<Int4>(oC.z) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.z = As<Float4>(As<Int4>(oC.z) & *Pointer<Int4>(constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.z = As<Float4>(As<Int4>(oC.z) | As<Int4>(masked));
}
oC.z = As<Float4>(As<Int4>(oC.z) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX2X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskX2X) + xMask * 16, 16));
oC.z = As<Float4>(As<Int4>(oC.z) & *Pointer<Int4>(constants + OFFSET(Constants,maskX2X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskX2X) + xMask * 16, 16));
oC.z = As<Float4>(As<Int4>(oC.z) | As<Int4>(value));
*Pointer<Float4>(buffer, 16) = oC.z;
}
......@@ -2301,13 +2289,13 @@ namespace sw
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
oC.w = As<Float4>(As<Int4>(oC.w) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.w = As<Float4>(As<Int4>(oC.w) & *Pointer<Int4>(constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
masked = As<Float4>(As<Int4>(masked) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskD4X[rgbaWriteMask][0])));
oC.w = As<Float4>(As<Int4>(oC.w) | As<Int4>(masked));
}
oC.w = As<Float4>(As<Int4>(oC.w) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX3X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(r.constants + OFFSET(Constants,invMaskX3X) + xMask * 16, 16));
oC.w = As<Float4>(As<Int4>(oC.w) & *Pointer<Int4>(constants + OFFSET(Constants,maskX3X) + xMask * 16, 16));
value = As<Float4>(As<Int4>(value) & *Pointer<Int4>(constants + OFFSET(Constants,invMaskX3X) + xMask * 16, 16));
oC.w = As<Float4>(As<Int4>(oC.w) | As<Int4>(value));
*Pointer<Float4>(buffer + 16, 16) = oC.w;
}
......@@ -2322,18 +2310,18 @@ namespace sw
return UShort4(cf * Float4(0xFFFF), saturate);
}
void PixelRoutine::sRGBtoLinear16_12_16(Registers &r, Vector4s &c)
void PixelRoutine::sRGBtoLinear16_12_16(Vector4s &c)
{
c.x = As<UShort4>(c.x) >> 4;
c.y = As<UShort4>(c.y) >> 4;
c.z = As<UShort4>(c.z) >> 4;
sRGBtoLinear12_16(r, c);
sRGBtoLinear12_16(c);
}
void PixelRoutine::sRGBtoLinear12_16(Registers &r, Vector4s &c)
void PixelRoutine::sRGBtoLinear12_16(Vector4s &c)
{
Pointer<Byte> LUT = r.constants + OFFSET(Constants,sRGBtoLinear12_16);
Pointer<Byte> LUT = constants + OFFSET(Constants,sRGBtoLinear12_16);
c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 0))), 0);
c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 1))), 1);
......@@ -2351,18 +2339,18 @@ namespace sw
c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 3))), 3);
}
void PixelRoutine::linearToSRGB16_12_16(Registers &r, Vector4s &c)
void PixelRoutine::linearToSRGB16_12_16(Vector4s &c)
{
c.x = As<UShort4>(c.x) >> 4;
c.y = As<UShort4>(c.y) >> 4;
c.z = As<UShort4>(c.z) >> 4;
linearToSRGB12_16(r, c);
linearToSRGB12_16(c);
}
void PixelRoutine::linearToSRGB12_16(Registers &r, Vector4s &c)
void PixelRoutine::linearToSRGB12_16(Vector4s &c)
{
Pointer<Byte> LUT = r.constants + OFFSET(Constants,linearToSRGB12_16);
Pointer<Byte> LUT = constants + OFFSET(Constants,linearToSRGB12_16);
c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 0))), 0);
c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 1))), 1);
......
src/Shader/PixelRoutine.hpp
......@@ -29,68 +29,62 @@ namespace sw
virtual ~PixelRoutine();
protected:
struct Registers : public QuadRasterizer::Registers
{
Registers(const PixelShader *shader);
Float4 z[4]; // Multisampled z
Float4 w; // Used as is
Float4 rhw; // Reciprocal w
Float4 z[4]; // Multisampled z
Float4 w; // Used as is
Float4 rhw; // Reciprocal w
RegisterArray<10> v; // Varying registers
// Depth output
Float4 oDepth;
};
RegisterArray<10> v; // Varying registers
// Depth output
Float4 oDepth;
typedef Shader::SourceParameter Src;
typedef Shader::DestinationParameter Dst;
virtual void setBuiltins(Registers &r, Int &x, Int &y, Float4(&z)[4], Float4 &w) = 0;
virtual void applyShader(Registers &r, Int cMask[4]) = 0;
virtual Bool alphaTest(Registers &r, Int cMask[4]) = 0;
virtual void rasterOperation(Registers &r, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]) = 0;
virtual void setBuiltins(Int &x, Int &y, Float4(&z)[4], Float4 &w) = 0;
virtual void applyShader(Int cMask[4]) = 0;
virtual Bool alphaTest(Int cMask[4]) = 0;
virtual void rasterOperation(Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]) = 0;
virtual void quad(QuadRasterizer::Registers &r, Pointer<Byte> cBuffer[4], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y);
virtual void quad(Pointer<Byte> cBuffer[4], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y);
void alphaTest(Registers &r, Int &aMask, Short4 &alpha);
void alphaToCoverage(Registers &r, Int cMask[4], Float4 &alpha);
void fogBlend(Registers &r, Vector4f &c0, Float4 &fog);
void pixelFog(Registers &r, Float4 &visibility);
void alphaTest(Int &aMask, Short4 &alpha);
void alphaToCoverage(Int cMask[4], Float4 &alpha);
void fogBlend(Vector4f &c0, Float4 &fog);
void pixelFog(Float4 &visibility);
// Raster operations
void alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x);
void logicOperation(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x);
void writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &i, Vector4s &current, Int &sMask, Int &zMask, Int &cMask);
void alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4f &oC, Int &x);
void writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &i, Vector4f &oC, Int &sMask, Int &zMask, Int &cMask);
void alphaBlend(int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x);
void logicOperation(int index, Pointer<Byte> &cBuffer, Vector4s &current, Int &x);
void writeColor(int index, Pointer<Byte> &cBuffer, Int &i, Vector4s &current, Int &sMask, Int &zMask, Int &cMask);
void alphaBlend(int index, Pointer<Byte> &cBuffer, Vector4f &oC, Int &x);
void writeColor(int index, Pointer<Byte> &cBuffer, Int &i, Vector4f &oC, Int &sMask, Int &zMask, Int &cMask);
UShort4 convertFixed16(Float4 &cf, bool saturate = true);
void linearToSRGB12_16(Registers &r, Vector4s &c);
void linearToSRGB12_16(Vector4s &c);
SamplerCore *sampler[TEXTURE_IMAGE_UNITS];
private:
Float4 interpolateCentroid(Float4 &x, Float4 &y, Float4 &rhw, Pointer<Byte> planeEquation, bool flat, bool perspective);
void stencilTest(Registers &r, Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &cMask);
void stencilTest(Registers &r, Byte8 &value, StencilCompareMode stencilCompareMode, bool CCW);
void stencilOperation(Registers &r, Byte8 &newValue, Byte8 &bufferValue, StencilOperation stencilPassOperation, StencilOperation stencilZFailOperation, StencilOperation stencilFailOperation, bool CCW, Int &zMask, Int &sMask);
void stencilOperation(Registers &r, Byte8 &output, Byte8 &bufferValue, StencilOperation operation, bool CCW);
Bool depthTest(Registers &r, Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &sMask, Int &zMask, Int &cMask);
void stencilTest(Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &cMask);
void stencilTest(Byte8 &value, StencilCompareMode stencilCompareMode, bool CCW);
void stencilOperation(Byte8 &newValue, Byte8 &bufferValue, StencilOperation stencilPassOperation, StencilOperation stencilZFailOperation, StencilOperation stencilFailOperation, bool CCW, Int &zMask, Int &sMask);
void stencilOperation(Byte8 &output, Byte8 &bufferValue, StencilOperation operation, bool CCW);
Bool depthTest(Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &sMask, Int &zMask, Int &cMask);
// Raster operations
void blendFactor(Registers &r, const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorActive);
void blendFactorAlpha(Registers &r, const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorAlphaActive);
void readPixel(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Vector4s &pixel);
void blendFactor(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorActive);
void blendFactorAlpha(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorAlphaActive);
void writeStencil(Registers &r, Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &zMask, Int &cMask);
void writeDepth(Registers &r, Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &zMask);
void sRGBtoLinear16_12_16(Registers &r, Vector4s &c);
void sRGBtoLinear12_16(Registers &r, Vector4s &c);
void linearToSRGB16_12_16(Registers &r, Vector4s &c);
void blendFactor(const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorActive);
void blendFactorAlpha(const Vector4s &blendFactor, const Vector4s &current, const Vector4s &pixel, BlendFactor blendFactorAlphaActive);
void readPixel(int index, Pointer<Byte> &cBuffer, Int &x, Vector4s &pixel);
void blendFactor(const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorActive);
void blendFactorAlpha(const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, BlendFactor blendFactorAlphaActive);
void writeStencil(Pointer<Byte> &sBuffer, int q, Int &x, Int &sMask, Int &zMask, Int &cMask);
void writeDepth(Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &zMask);
void sRGBtoLinear16_12_16(Vector4s &c);
void sRGBtoLinear12_16(Vector4s &c);
void linearToSRGB16_12_16(Vector4s &c);
Float4 sRGBtoLinear(const Float4 &x);
bool colorUsed();
......
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