Update SwiftShader to April code dump.
April code dump from Transgaming. Adds new shader compiler.
diff --git a/src/Shader/PixelRoutine.cpp b/src/Shader/PixelRoutine.cpp
index f1ac5be..e1b735d 100644
--- a/src/Shader/PixelRoutine.cpp
+++ b/src/Shader/PixelRoutine.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -12,7 +12,6 @@
#include "PixelRoutine.hpp"
#include "Renderer.hpp"
-#include "PixelShader.hpp"
#include "QuadRasterizer.hpp"
#include "Surface.hpp"
#include "Primitive.hpp"
@@ -30,17 +29,21 @@
extern bool complementaryDepthBuffer;
extern bool postBlendSRGB;
extern bool exactColorRounding;
+ extern bool booleanFaceRegister;
+ extern bool halfIntegerCoordinates; // Pixel centers are not at integer coordinates
+ extern bool fullPixelPositionRegister;
- PixelRoutine::PixelRoutine(const PixelProcessor::State &state, const PixelShader *pixelShader) : Rasterizer(state), pixelShader(pixelShader)
+ PixelRoutine::PixelRoutine(const PixelProcessor::State &state, const PixelShader *shader) : Rasterizer(state), shader(shader)
{
perturbate = false;
luminance = false;
previousScaling = false;
- returns = false;
ifDepth = 0;
loopRepDepth = 0;
breakDepth = 0;
+ currentLabel = -1;
+ whileTest = false;
for(int i = 0; i < 2048; i++)
{
@@ -68,7 +71,7 @@
}
const bool earlyDepthTest = !state.depthOverride && !state.alphaTestActive();
- const bool integerPipeline = pixelShaderVersion() <= 0x0104;
+ const bool integerPipeline = shaderVersion() <= 0x0104;
Int zMask[4]; // Depth mask
Int sMask[4]; // Stencil mask
@@ -86,18 +89,19 @@
Float4 f;
- Color4i ¤t = r.ri[0];
- Color4i &diffuse = r.vi[0];
- Color4i &specular = r.vi[1];
+ Vector4i ¤t = r.ri[0];
+ Vector4i &diffuse = r.vi[0];
+ Vector4i &specular = r.vi[1];
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 yyyy = Float4(Float(y)) + *Pointer<Float4>(r.primitive + OFFSET(Primitive,yQuad), 16);
- if(state.depthTestActive || state.pixelFogActive())
+ if(interpolateZ())
{
for(unsigned int q = 0; q < state.multiSample; q++)
{
@@ -151,9 +155,10 @@
YYYY += yyyy;
}
- if(state.perspective)
+ if(interpolateW())
{
- rhw = reciprocal(interpolate(xxxx, r.Dw, rhw, r.primitive + OFFSET(Primitive,w), false, false));
+ w = interpolate(xxxx, r.Dw, rhw, r.primitive + OFFSET(Primitive,w), false, false);
+ rhw = reciprocal(w);
if(state.centroid)
{
@@ -165,27 +170,15 @@
{
for(int component = 0; component < 4; component++)
{
- Array<Float4> *pv;
-
- switch(component)
- {
- case 0: pv = &r.vx; break;
- case 1: pv = &r.vy; break;
- case 2: pv = &r.vz; break;
- case 3: pv = &r.vw; break;
- }
-
- Array<Float4> &v = *pv;
-
if(state.interpolant[interpolant].component & (1 << component))
{
if(!state.interpolant[interpolant].centroid)
{
- v[interpolant] = interpolate(xxxx, r.Dv[interpolant][component], rhw, r.primitive + OFFSET(Primitive,V[interpolant][component]), state.interpolant[interpolant].flat & (1 << component), state.perspective);
+ r.vf[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);
}
else
{
- v[interpolant] = interpolateCentroid(XXXX, YYYY, rhwCentroid, r.primitive + OFFSET(Primitive,V[interpolant][component]), state.interpolant[interpolant].flat & (1 << component), state.perspective);
+ r.vf[interpolant][component] = interpolateCentroid(XXXX, YYYY, rhwCentroid, r.primitive + OFFSET(Primitive,V[interpolant][component]), (state.interpolant[interpolant].flat & (1 << component)) != 0, state.perspective);
}
}
}
@@ -197,19 +190,19 @@
case 0:
break;
case 1:
- rcp = reciprocal(Float4(r.vy[interpolant]));
- r.vx[interpolant] = r.vx[interpolant] * rcp;
+ rcp = reciprocal(r.vf[interpolant].y);
+ r.vf[interpolant].x = r.vf[interpolant].x * rcp;
break;
case 2:
- rcp = reciprocal(Float4(r.vz[interpolant]));
- r.vx[interpolant] = r.vx[interpolant] * rcp;
- r.vy[interpolant] = r.vy[interpolant] * rcp;
+ rcp = reciprocal(r.vf[interpolant].z);
+ r.vf[interpolant].x = r.vf[interpolant].x * rcp;
+ r.vf[interpolant].y = r.vf[interpolant].y * rcp;
break;
case 3:
- rcp = reciprocal(Float4(r.vw[interpolant]));
- r.vx[interpolant] = r.vx[interpolant] * rcp;
- r.vy[interpolant] = r.vy[interpolant] * rcp;
- r.vz[interpolant] = r.vz[interpolant] * rcp;
+ rcp = reciprocal(r.vf[interpolant].w);
+ r.vf[interpolant].x = r.vf[interpolant].x * rcp;
+ r.vf[interpolant].y = r.vf[interpolant].y * rcp;
+ r.vf[interpolant].z = r.vf[interpolant].z * rcp;
break;
}
}
@@ -221,32 +214,47 @@
if(integerPipeline)
{
- if(state.color[0].component & 0x1) diffuse.x = convertFixed12(Float4(r.vx[0])); else diffuse.x = Short4(0x1000, 0x1000, 0x1000, 0x1000);
- if(state.color[0].component & 0x2) diffuse.y = convertFixed12(Float4(r.vy[0])); else diffuse.y = Short4(0x1000, 0x1000, 0x1000, 0x1000);
- if(state.color[0].component & 0x4) diffuse.z = convertFixed12(Float4(r.vz[0])); else diffuse.z = Short4(0x1000, 0x1000, 0x1000, 0x1000);
- if(state.color[0].component & 0x8) diffuse.w = convertFixed12(Float4(r.vw[0])); else diffuse.w = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ if(state.color[0].component & 0x1) diffuse.x = convertFixed12(r.vf[0].x); else diffuse.x = Short4(0x1000);
+ if(state.color[0].component & 0x2) diffuse.y = convertFixed12(r.vf[0].y); else diffuse.y = Short4(0x1000);
+ if(state.color[0].component & 0x4) diffuse.z = convertFixed12(r.vf[0].z); else diffuse.z = Short4(0x1000);
+ if(state.color[0].component & 0x8) diffuse.w = convertFixed12(r.vf[0].w); else diffuse.w = Short4(0x1000);
- if(state.color[1].component & 0x1) specular.x = convertFixed12(Float4(r.vx[1])); else specular.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- if(state.color[1].component & 0x2) specular.y = convertFixed12(Float4(r.vy[1])); else specular.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- if(state.color[1].component & 0x4) specular.z = convertFixed12(Float4(r.vz[1])); else specular.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- if(state.color[1].component & 0x8) specular.w = convertFixed12(Float4(r.vw[1])); else specular.w = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ if(state.color[1].component & 0x1) specular.x = convertFixed12(r.vf[1].x); else specular.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ if(state.color[1].component & 0x2) specular.y = convertFixed12(r.vf[1].y); else specular.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ if(state.color[1].component & 0x4) specular.z = convertFixed12(r.vf[1].z); else specular.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ if(state.color[1].component & 0x8) specular.w = convertFixed12(r.vf[1].w); else specular.w = Short4(0x0000, 0x0000, 0x0000, 0x0000);
}
- else if(pixelShaderVersion() >= 0x0300)
+ else if(shaderVersion() >= 0x0300)
{
- if(pixelShader->vPosDeclared)
+ if(shader->vPosDeclared)
{
- r.vPos.x = Float4(Float(x)) + Float4(0, 1, 0, 1);
- r.vPos.y = Float4(Float(y)) + Float4(0, 0, 1, 1);
+ if(!halfIntegerCoordinates)
+ {
+ r.vPos.x = Float4(Float(x)) + Float4(0, 1, 0, 1);
+ r.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);
+ }
+
+ if(fullPixelPositionRegister)
+ {
+ r.vPos.z = z[0]; // FIXME: Centroid?
+ r.vPos.w = w; // FIXME: Centroid?
+ }
}
- if(pixelShader->vFaceDeclared)
+ if(shader->vFaceDeclared)
{
Float4 area = *Pointer<Float>(r.primitive + OFFSET(Primitive,area));
-
- r.vFace.x = area;
- r.vFace.y = area;
- r.vFace.z = area;
- r.vFace.w = area;
+ Float4 face = booleanFaceRegister ? As<Float4>(CmpNLT(area, Float4(0.0f))) : area;
+
+ r.vFace.x = face;
+ r.vFace.y = face;
+ r.vFace.z = face;
+ r.vFace.w = face;
}
}
@@ -262,11 +270,11 @@
Long shaderTime = Ticks();
#endif
- if(pixelShader)
+ if(shader)
{
- // pixelShader->print("PixelShader-%0.16llX.txt", state.shaderHash);
+ // shader->print("PixelShader-%0.8X.txt", state.shaderID);
- if(pixelShader->getVersion() <= 0x0104)
+ if(shader->getVersion() <= 0x0104)
{
ps_1_x(r, cMask);
}
@@ -278,7 +286,7 @@
else
{
current = diffuse;
- Color4i temp(0x0000, 0x0000, 0x0000, 0x0000);
+ Vector4i temp(0x0000, 0x0000, 0x0000, 0x0000);
for(int stage = 0; stage < 8; stage++)
{
@@ -287,7 +295,7 @@
break;
}
- Color4i texture;
+ Vector4i texture;
if(state.textureStage[stage].usesTexture)
{
@@ -306,10 +314,10 @@
if(integerPipeline)
{
- current.r = Min(current.r, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.r = Max(current.r, Short4(0x0000, 0x0000, 0x0000, 0x0000));
- current.g = Min(current.g, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.g = Max(current.g, Short4(0x0000, 0x0000, 0x0000, 0x0000));
- current.b = Min(current.b, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.b = Max(current.b, Short4(0x0000, 0x0000, 0x0000, 0x0000));
- current.a = Min(current.a, Short4(0x0FFF, 0x0FFF, 0x0FFF, 0x0FFF)); current.a = Max(current.a, 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));
alphaPass = alphaTest(r, cMask, current);
}
@@ -320,7 +328,7 @@
alphaPass = alphaTest(r, cMask, r.oC[0]);
}
- if((pixelShader && pixelShader->containsTexkill()) || state.alphaTestActive())
+ if((shader && shader->containsKill()) || state.alphaTestActive())
{
for(unsigned int q = 0; q < state.multiSample; q++)
{
@@ -528,11 +536,11 @@
Float4 Z = z;
- if(pixelShader && pixelShader->depthOverride())
+ if(shader && shader->depthOverride())
{
if(complementaryDepthBuffer)
{
- Z = Float4(1, 1, 1, 1) - r.oDepth;
+ Z = Float4(1.0f) - r.oDepth;
}
else
{
@@ -655,15 +663,15 @@
return zMask != 0;
}
- void PixelRoutine::blendTexture(Registers &r, Color4i ¤t, Color4i &temp, Color4i &texture, int stage)
+ void PixelRoutine::blendTexture(Registers &r, Vector4i ¤t, Vector4i &temp, Vector4i &texture, int stage)
{
- Color4i *arg1;
- Color4i *arg2;
- Color4i *arg3;
- Color4i res;
+ Vector4i *arg1;
+ Vector4i *arg2;
+ Vector4i *arg3;
+ Vector4i res;
- Color4i constant;
- Color4i tfactor;
+ Vector4i constant;
+ Vector4i tfactor;
const TextureStage::State &textureStage = state.textureStage[stage];
@@ -674,10 +682,10 @@
textureStage.thirdArgument == TextureStage::SOURCE_CONSTANT ||
textureStage.thirdArgumentAlpha == TextureStage::SOURCE_CONSTANT)
{
- constant.r = *Pointer<Short4>(r.data + OFFSET(DrawData,textureStage[stage].constantColor4[0]));
- constant.g = *Pointer<Short4>(r.data + OFFSET(DrawData,textureStage[stage].constantColor4[1]));
- constant.b = *Pointer<Short4>(r.data + OFFSET(DrawData,textureStage[stage].constantColor4[2]));
- constant.a = *Pointer<Short4>(r.data + OFFSET(DrawData,textureStage[stage].constantColor4[3]));
+ 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]));
}
if(textureStage.firstArgument == TextureStage::SOURCE_TFACTOR ||
@@ -687,10 +695,10 @@
textureStage.thirdArgument == TextureStage::SOURCE_TFACTOR ||
textureStage.thirdArgumentAlpha == TextureStage::SOURCE_TFACTOR)
{
- tfactor.r = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[0]));
- tfactor.g = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[1]));
- tfactor.b = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[2]));
- tfactor.a = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[3]));
+ 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]));
}
// Premodulate
@@ -698,22 +706,22 @@
{
if(state.textureStage[stage - 1].stageOperation == TextureStage::STAGE_PREMODULATE)
{
- current.r = MulHigh(current.r, texture.r) << 4;
- current.g = MulHigh(current.g, texture.g) << 4;
- current.b = MulHigh(current.b, texture.b) << 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)
{
- current.a = MulHigh(current.a, texture.a) << 4;
+ current.w = MulHigh(current.w, texture.w) << 4;
}
}
if(luminance)
{
- texture.r = MulHigh(texture.r, r.L) << 4;
- texture.g = MulHigh(texture.g, r.L) << 4;
- texture.b = MulHigh(texture.b, r.L) << 4;
+ texture.x = MulHigh(texture.x, r.L) << 4;
+ texture.y = MulHigh(texture.y, r.L) << 4;
+ texture.z = MulHigh(texture.z, r.L) << 4;
luminance = false;
}
@@ -757,9 +765,9 @@
ASSERT(false);
}
- Color4i mod1;
- Color4i mod2;
- Color4i mod3;
+ Vector4i mod1;
+ Vector4i mod2;
+ Vector4i mod3;
switch(textureStage.firstModifier)
{
@@ -767,30 +775,30 @@
break;
case TextureStage::MODIFIER_INVCOLOR:
{
- mod1.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->r);
- mod1.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->g);
- mod1.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->b);
- mod1.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->a);
+ mod1.x = SubSat(Short4(0x1000), arg1->x);
+ mod1.y = SubSat(Short4(0x1000), arg1->y);
+ mod1.z = SubSat(Short4(0x1000), arg1->z);
+ mod1.w = SubSat(Short4(0x1000), arg1->w);
arg1 = &mod1;
}
break;
case TextureStage::MODIFIER_ALPHA:
{
- mod1.r = arg1->a;
- mod1.g = arg1->a;
- mod1.b = arg1->a;
- mod1.a = arg1->a;
+ mod1.x = arg1->w;
+ mod1.y = arg1->w;
+ mod1.z = arg1->w;
+ mod1.w = arg1->w;
arg1 = &mod1;
}
break;
case TextureStage::MODIFIER_INVALPHA:
{
- mod1.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->a);
- mod1.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->a);
- mod1.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->a);
- mod1.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->a);
+ mod1.x = SubSat(Short4(0x1000), arg1->w);
+ mod1.y = SubSat(Short4(0x1000), arg1->w);
+ mod1.z = SubSat(Short4(0x1000), arg1->w);
+ mod1.w = SubSat(Short4(0x1000), arg1->w);
arg1 = &mod1;
}
@@ -805,30 +813,30 @@
break;
case TextureStage::MODIFIER_INVCOLOR:
{
- mod2.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->r);
- mod2.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->g);
- mod2.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->b);
- mod2.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->a);
+ mod2.x = SubSat(Short4(0x1000), arg2->x);
+ mod2.y = SubSat(Short4(0x1000), arg2->y);
+ mod2.z = SubSat(Short4(0x1000), arg2->z);
+ mod2.w = SubSat(Short4(0x1000), arg2->w);
arg2 = &mod2;
}
break;
case TextureStage::MODIFIER_ALPHA:
{
- mod2.r = arg2->a;
- mod2.g = arg2->a;
- mod2.b = arg2->a;
- mod2.a = arg2->a;
+ mod2.x = arg2->w;
+ mod2.y = arg2->w;
+ mod2.z = arg2->w;
+ mod2.w = arg2->w;
arg2 = &mod2;
}
break;
case TextureStage::MODIFIER_INVALPHA:
{
- mod2.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->a);
- mod2.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->a);
- mod2.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->a);
- mod2.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->a);
+ mod2.x = SubSat(Short4(0x1000), arg2->w);
+ mod2.y = SubSat(Short4(0x1000), arg2->w);
+ mod2.z = SubSat(Short4(0x1000), arg2->w);
+ mod2.w = SubSat(Short4(0x1000), arg2->w);
arg2 = &mod2;
}
@@ -843,30 +851,30 @@
break;
case TextureStage::MODIFIER_INVCOLOR:
{
- mod3.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->r);
- mod3.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->g);
- mod3.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->b);
- mod3.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->a);
+ mod3.x = SubSat(Short4(0x1000), arg3->x);
+ mod3.y = SubSat(Short4(0x1000), arg3->y);
+ mod3.z = SubSat(Short4(0x1000), arg3->z);
+ mod3.w = SubSat(Short4(0x1000), arg3->w);
arg3 = &mod3;
}
break;
case TextureStage::MODIFIER_ALPHA:
{
- mod3.r = arg3->a;
- mod3.g = arg3->a;
- mod3.b = arg3->a;
- mod3.a = arg3->a;
+ mod3.x = arg3->w;
+ mod3.y = arg3->w;
+ mod3.z = arg3->w;
+ mod3.w = arg3->w;
arg3 = &mod3;
}
break;
case TextureStage::MODIFIER_INVALPHA:
{
- mod3.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->a);
- mod3.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->a);
- mod3.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->a);
- mod3.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->a);
+ mod3.x = SubSat(Short4(0x1000), arg3->w);
+ mod3.y = SubSat(Short4(0x1000), arg3->w);
+ mod3.z = SubSat(Short4(0x1000), arg3->w);
+ mod3.w = SubSat(Short4(0x1000), arg3->w);
arg3 = &mod3;
}
@@ -881,211 +889,211 @@
break;
case TextureStage::STAGE_SELECTARG1: // Arg1
{
- res.r = arg1->r;
- res.g = arg1->g;
- res.b = arg1->b;
+ res.x = arg1->x;
+ res.y = arg1->y;
+ res.z = arg1->z;
}
break;
case TextureStage::STAGE_SELECTARG2: // Arg2
{
- res.r = arg2->r;
- res.g = arg2->g;
- res.b = arg2->b;
+ res.x = arg2->x;
+ res.y = arg2->y;
+ res.z = arg2->z;
}
break;
case TextureStage::STAGE_SELECTARG3: // Arg3
{
- res.r = arg3->r;
- res.g = arg3->g;
- res.b = arg3->b;
+ res.x = arg3->x;
+ res.y = arg3->y;
+ res.z = arg3->z;
}
break;
case TextureStage::STAGE_MODULATE: // Arg1 * Arg2
{
- res.r = MulHigh(arg1->r, arg2->r) << 4;
- res.g = MulHigh(arg1->g, arg2->g) << 4;
- res.b = MulHigh(arg1->b, arg2->b) << 4;
+ res.x = MulHigh(arg1->x, arg2->x) << 4;
+ res.y = MulHigh(arg1->y, arg2->y) << 4;
+ res.z = MulHigh(arg1->z, arg2->z) << 4;
}
break;
case TextureStage::STAGE_MODULATE2X: // Arg1 * Arg2 * 2
{
- res.r = MulHigh(arg1->r, arg2->r) << 5;
- res.g = MulHigh(arg1->g, arg2->g) << 5;
- res.b = MulHigh(arg1->b, arg2->b) << 5;
+ res.x = MulHigh(arg1->x, arg2->x) << 5;
+ res.y = MulHigh(arg1->y, arg2->y) << 5;
+ res.z = MulHigh(arg1->z, arg2->z) << 5;
}
break;
case TextureStage::STAGE_MODULATE4X: // Arg1 * Arg2 * 4
{
- res.r = MulHigh(arg1->r, arg2->r) << 6;
- res.g = MulHigh(arg1->g, arg2->g) << 6;
- res.b = MulHigh(arg1->b, arg2->b) << 6;
+ res.x = MulHigh(arg1->x, arg2->x) << 6;
+ res.y = MulHigh(arg1->y, arg2->y) << 6;
+ res.z = MulHigh(arg1->z, arg2->z) << 6;
}
break;
case TextureStage::STAGE_ADD: // Arg1 + Arg2
{
- res.r = AddSat(arg1->r, arg2->r);
- res.g = AddSat(arg1->g, arg2->g);
- res.b = AddSat(arg1->b, arg2->b);
+ res.x = AddSat(arg1->x, arg2->x);
+ res.y = AddSat(arg1->y, arg2->y);
+ res.z = AddSat(arg1->z, arg2->z);
}
break;
case TextureStage::STAGE_ADDSIGNED: // Arg1 + Arg2 - 0.5
{
- res.r = AddSat(arg1->r, arg2->r);
- res.g = AddSat(arg1->g, arg2->g);
- res.b = AddSat(arg1->b, arg2->b);
+ res.x = AddSat(arg1->x, arg2->x);
+ res.y = AddSat(arg1->y, arg2->y);
+ res.z = AddSat(arg1->z, arg2->z);
- res.r = SubSat(res.r, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- res.g = SubSat(res.g, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- res.b = SubSat(res.b, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.x = SubSat(res.x, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.y = SubSat(res.y, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.z = SubSat(res.z, Short4(0x0800, 0x0800, 0x0800, 0x0800));
}
break;
case TextureStage::STAGE_ADDSIGNED2X: // (Arg1 + Arg2 - 0.5) << 1
{
- res.r = AddSat(arg1->r, arg2->r);
- res.g = AddSat(arg1->g, arg2->g);
- res.b = AddSat(arg1->b, arg2->b);
+ res.x = AddSat(arg1->x, arg2->x);
+ res.y = AddSat(arg1->y, arg2->y);
+ res.z = AddSat(arg1->z, arg2->z);
- res.r = SubSat(res.r, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- res.g = SubSat(res.g, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- res.b = SubSat(res.b, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.x = SubSat(res.x, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.y = SubSat(res.y, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.z = SubSat(res.z, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- res.r = AddSat(res.r, res.r);
- res.g = AddSat(res.g, res.g);
- res.b = AddSat(res.b, res.b);
+ res.x = AddSat(res.x, res.x);
+ res.y = AddSat(res.y, res.y);
+ res.z = AddSat(res.z, res.z);
}
break;
case TextureStage::STAGE_SUBTRACT: // Arg1 - Arg2
{
- res.r = SubSat(arg1->r, arg2->r);
- res.g = SubSat(arg1->g, arg2->g);
- res.b = SubSat(arg1->b, arg2->b);
+ res.x = SubSat(arg1->x, arg2->x);
+ res.y = SubSat(arg1->y, arg2->y);
+ res.z = SubSat(arg1->z, arg2->z);
}
break;
case TextureStage::STAGE_ADDSMOOTH: // Arg1 + Arg2 - Arg1 * Arg2
{
Short4 tmp;
- tmp = MulHigh(arg1->r, arg2->r) << 4; res.r = AddSat(arg1->r, arg2->r); res.r = SubSat(res.r, tmp);
- tmp = MulHigh(arg1->g, arg2->g) << 4; res.g = AddSat(arg1->g, arg2->g); res.g = SubSat(res.g, tmp);
- tmp = MulHigh(arg1->b, arg2->b) << 4; res.b = AddSat(arg1->b, arg2->b); res.b = SubSat(res.b, tmp);
+ tmp = MulHigh(arg1->x, arg2->x) << 4; res.x = AddSat(arg1->x, arg2->x); res.x = SubSat(res.x, tmp);
+ tmp = MulHigh(arg1->y, arg2->y) << 4; res.y = AddSat(arg1->y, arg2->y); res.y = SubSat(res.y, tmp);
+ tmp = MulHigh(arg1->z, arg2->z) << 4; res.z = AddSat(arg1->z, arg2->z); res.z = SubSat(res.z, tmp);
}
break;
case TextureStage::STAGE_MULTIPLYADD: // Arg3 + Arg1 * Arg2
{
- res.r = MulHigh(arg1->r, arg2->r) << 4; res.r = AddSat(res.r, arg3->r);
- res.g = MulHigh(arg1->g, arg2->g) << 4; res.g = AddSat(res.g, arg3->g);
- res.b = MulHigh(arg1->b, arg2->b) << 4; res.b = AddSat(res.b, arg3->b);
+ res.x = MulHigh(arg1->x, arg2->x) << 4; res.x = AddSat(res.x, arg3->x);
+ res.y = MulHigh(arg1->y, arg2->y) << 4; res.y = AddSat(res.y, arg3->y);
+ res.z = MulHigh(arg1->z, arg2->z) << 4; res.z = AddSat(res.z, arg3->z);
}
break;
case TextureStage::STAGE_LERP: // Arg3 * (Arg1 - Arg2) + Arg2
{
- res.r = SubSat(arg1->r, arg2->r); res.r = MulHigh(res.r, arg3->r) << 4; res.r = AddSat(res.r, arg2->r);
- res.g = SubSat(arg1->g, arg2->g); res.g = MulHigh(res.g, arg3->g) << 4; res.g = AddSat(res.g, arg2->g);
- res.b = SubSat(arg1->b, arg2->b); res.b = MulHigh(res.b, arg3->b) << 4; res.b = AddSat(res.b, arg2->b);
+ res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, arg3->x) << 4; res.x = AddSat(res.x, arg2->x);
+ res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, arg3->y) << 4; res.y = AddSat(res.y, arg2->y);
+ res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, arg3->z) << 4; res.z = AddSat(res.z, arg2->z);
}
break;
- case TextureStage::STAGE_DOT3: // 2 * (Arg1.r - 0.5) * 2 * (Arg2.r - 0.5) + 2 * (Arg1.g - 0.5) * 2 * (Arg2.g - 0.5) + 2 * (Arg1.b - 0.5) * 2 * (Arg2.b - 0.5)
+ case TextureStage::STAGE_DOT3: // 2 * (Arg1.x - 0.5) * 2 * (Arg2.x - 0.5) + 2 * (Arg1.y - 0.5) * 2 * (Arg2.y - 0.5) + 2 * (Arg1.z - 0.5) * 2 * (Arg2.z - 0.5)
{
Short4 tmp;
- res.r = SubSat(arg1->r, Short4(0x0800, 0x0800, 0x0800, 0x0800)); tmp = SubSat(arg2->r, Short4(0x0800, 0x0800, 0x0800, 0x0800)); res.r = MulHigh(res.r, tmp);
- res.g = SubSat(arg1->g, Short4(0x0800, 0x0800, 0x0800, 0x0800)); tmp = SubSat(arg2->g, Short4(0x0800, 0x0800, 0x0800, 0x0800)); res.g = MulHigh(res.g, tmp);
- res.b = SubSat(arg1->b, Short4(0x0800, 0x0800, 0x0800, 0x0800)); tmp = SubSat(arg2->b, Short4(0x0800, 0x0800, 0x0800, 0x0800)); res.b = MulHigh(res.b, tmp);
+ res.x = SubSat(arg1->x, Short4(0x0800, 0x0800, 0x0800, 0x0800)); tmp = SubSat(arg2->x, Short4(0x0800, 0x0800, 0x0800, 0x0800)); res.x = MulHigh(res.x, tmp);
+ res.y = SubSat(arg1->y, Short4(0x0800, 0x0800, 0x0800, 0x0800)); tmp = SubSat(arg2->y, Short4(0x0800, 0x0800, 0x0800, 0x0800)); res.y = MulHigh(res.y, tmp);
+ res.z = SubSat(arg1->z, Short4(0x0800, 0x0800, 0x0800, 0x0800)); tmp = SubSat(arg2->z, Short4(0x0800, 0x0800, 0x0800, 0x0800)); res.z = MulHigh(res.z, tmp);
- res.r = res.r << 6;
- res.g = res.g << 6;
- res.b = res.b << 6;
+ res.x = res.x << 6;
+ res.y = res.y << 6;
+ res.z = res.z << 6;
- res.r = AddSat(res.r, res.g);
- res.r = AddSat(res.r, res.b);
+ res.x = AddSat(res.x, res.y);
+ res.x = AddSat(res.x, res.z);
// Clamp to [0, 1]
- res.r = Max(res.r, Short4(0x0000, 0x0000, 0x0000, 0x0000));
- res.r = Min(res.r, Short4(0x1000, 0x1000, 0x1000, 0x1000));
+ res.x = Max(res.x, Short4(0x0000, 0x0000, 0x0000, 0x0000));
+ res.x = Min(res.x, Short4(0x1000));
- res.g = res.r;
- res.b = res.r;
- res.a = res.r;
+ res.y = res.x;
+ res.z = res.x;
+ res.w = res.x;
}
break;
case TextureStage::STAGE_BLENDCURRENTALPHA: // Alpha * (Arg1 - Arg2) + Arg2
{
- res.r = SubSat(arg1->r, arg2->r); res.r = MulHigh(res.r, current.a) << 4; res.r = AddSat(res.r, arg2->r);
- res.g = SubSat(arg1->g, arg2->g); res.g = MulHigh(res.g, current.a) << 4; res.g = AddSat(res.g, arg2->g);
- res.b = SubSat(arg1->b, arg2->b); res.b = MulHigh(res.b, current.a) << 4; res.b = AddSat(res.b, arg2->b);
+ 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.r = SubSat(arg1->r, arg2->r); res.r = MulHigh(res.r, r.diffuse.a) << 4; res.r = AddSat(res.r, arg2->r);
- res.g = SubSat(arg1->g, arg2->g); res.g = MulHigh(res.g, r.diffuse.a) << 4; res.g = AddSat(res.g, arg2->g);
- res.b = SubSat(arg1->b, arg2->b); res.b = MulHigh(res.b, r.diffuse.a) << 4; res.b = AddSat(res.b, arg2->b);
+ 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);
}
break;
case TextureStage::STAGE_BLENDFACTORALPHA: // Alpha * (Arg1 - Arg2) + Arg2
{
- res.r = SubSat(arg1->r, arg2->r); res.r = MulHigh(res.r, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[3]))) << 4; res.r = AddSat(res.r, arg2->r);
- res.g = SubSat(arg1->g, arg2->g); res.g = MulHigh(res.g, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[3]))) << 4; res.g = AddSat(res.g, arg2->g);
- res.b = SubSat(arg1->b, arg2->b); res.b = MulHigh(res.b, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[3]))) << 4; res.b = AddSat(res.b, arg2->b);
+ 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);
}
break;
case TextureStage::STAGE_BLENDTEXTUREALPHA: // Alpha * (Arg1 - Arg2) + Arg2
{
- res.r = SubSat(arg1->r, arg2->r); res.r = MulHigh(res.r, texture.a) << 4; res.r = AddSat(res.r, arg2->r);
- res.g = SubSat(arg1->g, arg2->g); res.g = MulHigh(res.g, texture.a) << 4; res.g = AddSat(res.g, arg2->g);
- res.b = SubSat(arg1->b, arg2->b); res.b = MulHigh(res.b, texture.a) << 4; res.b = AddSat(res.b, arg2->b);
+ res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, texture.w) << 4; res.x = AddSat(res.x, arg2->x);
+ res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, texture.w) << 4; res.y = AddSat(res.y, arg2->y);
+ res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, texture.w) << 4; res.z = AddSat(res.z, arg2->z);
}
break;
case TextureStage::STAGE_BLENDTEXTUREALPHAPM: // Arg1 + Arg2 * (1 - Alpha)
{
- res.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), texture.a); res.r = MulHigh(res.r, arg2->r) << 4; res.r = AddSat(res.r, arg1->r);
- res.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), texture.a); res.g = MulHigh(res.g, arg2->g) << 4; res.g = AddSat(res.g, arg1->g);
- res.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), texture.a); res.b = MulHigh(res.b, arg2->b) << 4; res.b = AddSat(res.b, arg1->b);
+ res.x = SubSat(Short4(0x1000), texture.w); res.x = MulHigh(res.x, arg2->x) << 4; res.x = AddSat(res.x, arg1->x);
+ res.y = SubSat(Short4(0x1000), texture.w); res.y = MulHigh(res.y, arg2->y) << 4; res.y = AddSat(res.y, arg1->y);
+ res.z = SubSat(Short4(0x1000), texture.w); res.z = MulHigh(res.z, arg2->z) << 4; res.z = AddSat(res.z, arg1->z);
}
break;
case TextureStage::STAGE_PREMODULATE:
{
- res.r = arg1->r;
- res.g = arg1->g;
- res.b = arg1->b;
+ res.x = arg1->x;
+ res.y = arg1->y;
+ res.z = arg1->z;
}
break;
- case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: // Arg1 + Arg1.a * Arg2
+ case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: // Arg1 + Arg1.w * Arg2
{
- res.r = MulHigh(arg1->a, arg2->r) << 4; res.r = AddSat(res.r, arg1->r);
- res.g = MulHigh(arg1->a, arg2->g) << 4; res.g = AddSat(res.g, arg1->g);
- res.b = MulHigh(arg1->a, arg2->b) << 4; res.b = AddSat(res.b, arg1->b);
+ res.x = MulHigh(arg1->w, arg2->x) << 4; res.x = AddSat(res.x, arg1->x);
+ res.y = MulHigh(arg1->w, arg2->y) << 4; res.y = AddSat(res.y, arg1->y);
+ res.z = MulHigh(arg1->w, arg2->z) << 4; res.z = AddSat(res.z, arg1->z);
}
break;
- case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: // Arg1 * Arg2 + Arg1.a
+ case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: // Arg1 * Arg2 + Arg1.w
{
- res.r = MulHigh(arg1->r, arg2->r) << 4; res.r = AddSat(res.r, arg1->a);
- res.g = MulHigh(arg1->g, arg2->g) << 4; res.g = AddSat(res.g, arg1->a);
- res.b = MulHigh(arg1->b, arg2->b) << 4; res.b = AddSat(res.b, arg1->a);
+ res.x = MulHigh(arg1->x, arg2->x) << 4; res.x = AddSat(res.x, arg1->w);
+ res.y = MulHigh(arg1->y, arg2->y) << 4; res.y = AddSat(res.y, arg1->w);
+ res.z = MulHigh(arg1->z, arg2->z) << 4; res.z = AddSat(res.z, arg1->w);
}
break;
- case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: // (1 - Arg1.a) * Arg2 + Arg1
+ case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: // (1 - Arg1.w) * Arg2 + Arg1
{
Short4 tmp;
- res.r = AddSat(arg1->r, arg2->r); tmp = MulHigh(arg1->a, arg2->r) << 4; res.r = SubSat(res.r, tmp);
- res.g = AddSat(arg1->g, arg2->g); tmp = MulHigh(arg1->a, arg2->g) << 4; res.g = SubSat(res.g, tmp);
- res.b = AddSat(arg1->b, arg2->b); tmp = MulHigh(arg1->a, arg2->b) << 4; res.b = SubSat(res.b, tmp);
+ res.x = AddSat(arg1->x, arg2->x); tmp = MulHigh(arg1->w, arg2->x) << 4; res.x = SubSat(res.x, tmp);
+ res.y = AddSat(arg1->y, arg2->y); tmp = MulHigh(arg1->w, arg2->y) << 4; res.y = SubSat(res.y, tmp);
+ res.z = AddSat(arg1->z, arg2->z); tmp = MulHigh(arg1->w, arg2->z) << 4; res.z = SubSat(res.z, tmp);
}
break;
- case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: // (1 - Arg1) * Arg2 + Arg1.a
+ case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: // (1 - Arg1) * Arg2 + Arg1.w
{
Short4 tmp;
- res.r = AddSat(arg1->a, arg2->r); tmp = MulHigh(arg1->r, arg2->r) << 4; res.r = SubSat(res.r, tmp);
- res.g = AddSat(arg1->a, arg2->g); tmp = MulHigh(arg1->g, arg2->g) << 4; res.g = SubSat(res.g, tmp);
- res.b = AddSat(arg1->a, arg2->b); tmp = MulHigh(arg1->b, arg2->b) << 4; res.b = SubSat(res.b, tmp);
+ res.x = AddSat(arg1->w, arg2->x); tmp = MulHigh(arg1->x, arg2->x) << 4; res.x = SubSat(res.x, tmp);
+ res.y = AddSat(arg1->w, arg2->y); tmp = MulHigh(arg1->y, arg2->y) << 4; res.y = SubSat(res.y, tmp);
+ res.z = AddSat(arg1->w, arg2->z); tmp = MulHigh(arg1->z, arg2->z) << 4; res.z = SubSat(res.z, tmp);
}
break;
case TextureStage::STAGE_BUMPENVMAP:
{
- r.du = Float4(texture.r) * Float4(1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0);
- r.dv = Float4(texture.g) * Float4(1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0);
+ r.du = Float4(texture.x) * Float4(1.0f / 0x0FE0);
+ r.dv = Float4(texture.y) * Float4(1.0f / 0x0FE0);
Float4 du2;
Float4 dv2;
@@ -1101,16 +1109,16 @@
perturbate = true;
- res.r = r.current.r;
- res.g = r.current.g;
- res.b = r.current.b;
- res.a = r.current.a;
+ res.x = r.current.x;
+ res.y = r.current.y;
+ res.z = r.current.z;
+ res.w = r.current.w;
}
break;
case TextureStage::STAGE_BUMPENVMAPLUMINANCE:
{
- r.du = Float4(texture.r) * Float4(1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0);
- r.dv = Float4(texture.g) * Float4(1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0, 1.0f / 0x0FE0);
+ r.du = Float4(texture.x) * Float4(1.0f / 0x0FE0);
+ r.dv = Float4(texture.y) * Float4(1.0f / 0x0FE0);
Float4 du2;
Float4 dv2;
@@ -1127,19 +1135,19 @@
perturbate = true;
- r.L = texture.b;
+ 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, 0x1000, 0x1000, 0x1000));
+ r.L = Min(r.L, Short4(0x1000));
luminance = true;
- res.r = r.current.r;
- res.g = r.current.g;
- res.b = r.current.b;
- res.a = r.current.a;
+ res.x = r.current.x;
+ res.y = r.current.y;
+ res.z = r.current.z;
+ res.w = r.current.w;
}
break;
default:
@@ -1193,7 +1201,7 @@
break;
case TextureStage::MODIFIER_INVCOLOR:
{
- mod1.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->a);
+ mod1.w = SubSat(Short4(0x1000), arg1->w);
arg1 = &mod1;
}
@@ -1205,7 +1213,7 @@
break;
case TextureStage::MODIFIER_INVALPHA:
{
- mod1.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg1->a);
+ mod1.w = SubSat(Short4(0x1000), arg1->w);
arg1 = &mod1;
}
@@ -1220,7 +1228,7 @@
break;
case TextureStage::MODIFIER_INVCOLOR:
{
- mod2.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->a);
+ mod2.w = SubSat(Short4(0x1000), arg2->w);
arg2 = &mod2;
}
@@ -1232,7 +1240,7 @@
break;
case TextureStage::MODIFIER_INVALPHA:
{
- mod2.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg2->a);
+ mod2.w = SubSat(Short4(0x1000), arg2->w);
arg2 = &mod2;
}
@@ -1247,7 +1255,7 @@
break;
case TextureStage::MODIFIER_INVCOLOR:
{
- mod3.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->a);
+ mod3.w = SubSat(Short4(0x1000), arg3->w);
arg3 = &mod3;
}
@@ -1259,7 +1267,7 @@
break;
case TextureStage::MODIFIER_INVALPHA:
{
- mod3.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), arg3->a);
+ mod3.w = SubSat(Short4(0x1000), arg3->w);
arg3 = &mod3;
}
@@ -1274,104 +1282,104 @@
break;
case TextureStage::STAGE_SELECTARG1: // Arg1
{
- res.a = arg1->a;
+ res.w = arg1->w;
}
break;
case TextureStage::STAGE_SELECTARG2: // Arg2
{
- res.a = arg2->a;
+ res.w = arg2->w;
}
break;
case TextureStage::STAGE_SELECTARG3: // Arg3
{
- res.a = arg3->a;
+ res.w = arg3->w;
}
break;
case TextureStage::STAGE_MODULATE: // Arg1 * Arg2
{
- res.a = MulHigh(arg1->a, arg2->a) << 4;
+ res.w = MulHigh(arg1->w, arg2->w) << 4;
}
break;
case TextureStage::STAGE_MODULATE2X: // Arg1 * Arg2 * 2
{
- res.a = MulHigh(arg1->a, arg2->a) << 5;
+ res.w = MulHigh(arg1->w, arg2->w) << 5;
}
break;
case TextureStage::STAGE_MODULATE4X: // Arg1 * Arg2 * 4
{
- res.a = MulHigh(arg1->a, arg2->a) << 6;
+ res.w = MulHigh(arg1->w, arg2->w) << 6;
}
break;
case TextureStage::STAGE_ADD: // Arg1 + Arg2
{
- res.a = AddSat(arg1->a, arg2->a);
+ res.w = AddSat(arg1->w, arg2->w);
}
break;
case TextureStage::STAGE_ADDSIGNED: // Arg1 + Arg2 - 0.5
{
- res.a = AddSat(arg1->a, arg2->a);
- res.a = SubSat(res.a, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.w = AddSat(arg1->w, arg2->w);
+ res.w = SubSat(res.w, Short4(0x0800, 0x0800, 0x0800, 0x0800));
}
break;
case TextureStage::STAGE_ADDSIGNED2X: // (Arg1 + Arg2 - 0.5) << 1
{
- res.a = AddSat(arg1->a, arg2->a);
- res.a = SubSat(res.a, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- res.a = AddSat(res.a, res.a);
+ res.w = AddSat(arg1->w, arg2->w);
+ res.w = SubSat(res.w, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ res.w = AddSat(res.w, res.w);
}
break;
case TextureStage::STAGE_SUBTRACT: // Arg1 - Arg2
{
- res.a = SubSat(arg1->a, arg2->a);
+ res.w = SubSat(arg1->w, arg2->w);
}
break;
case TextureStage::STAGE_ADDSMOOTH: // Arg1 + Arg2 - Arg1 * Arg2
{
Short4 tmp;
- tmp = MulHigh(arg1->a, arg2->a) << 4; res.a = AddSat(arg1->a, arg2->a); res.a = SubSat(res.a, tmp);
+ tmp = MulHigh(arg1->w, arg2->w) << 4; res.w = AddSat(arg1->w, arg2->w); res.w = SubSat(res.w, tmp);
}
break;
case TextureStage::STAGE_MULTIPLYADD: // Arg3 + Arg1 * Arg2
{
- res.a = MulHigh(arg1->a, arg2->a) << 4; res.a = AddSat(res.a, arg3->a);
+ res.w = MulHigh(arg1->w, arg2->w) << 4; res.w = AddSat(res.w, arg3->w);
}
break;
case TextureStage::STAGE_LERP: // Arg3 * (Arg1 - Arg2) + Arg2
{
- res.a = SubSat(arg1->a, arg2->a); res.a = MulHigh(res.a, arg3->a) << 4; res.a = AddSat(res.a, arg2->a);
+ res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, arg3->w) << 4; res.w = AddSat(res.w, arg2->w);
}
break;
case TextureStage::STAGE_DOT3:
break; // Already computed in color channel
case TextureStage::STAGE_BLENDCURRENTALPHA: // Alpha * (Arg1 - Arg2) + Arg2
{
- res.a = SubSat(arg1->a, arg2->a); res.a = MulHigh(res.a, current.a) << 4; res.a = AddSat(res.a, arg2->a);
+ 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.a = SubSat(arg1->a, arg2->a); res.a = MulHigh(res.a, r.diffuse.a) << 4; res.a = AddSat(res.a, arg2->a);
+ res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, r.diffuse.w) << 4; res.w = AddSat(res.w, arg2->w);
}
break;
case TextureStage::STAGE_BLENDFACTORALPHA:
{
- res.a = SubSat(arg1->a, arg2->a); res.a = MulHigh(res.a, *Pointer<Short4>(r.data + OFFSET(DrawData,factor.textureFactor4[3]))) << 4; res.a = AddSat(res.a, arg2->a);
+ 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);
}
break;
case TextureStage::STAGE_BLENDTEXTUREALPHA: // Arg1 * (Alpha) + Arg2 * (1 - Alpha)
{
- res.a = SubSat(arg1->a, arg2->a); res.a = MulHigh(res.a, texture.a) << 4; res.a = AddSat(res.a, arg2->a);
+ res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, texture.w) << 4; res.w = AddSat(res.w, arg2->w);
}
break;
case TextureStage::STAGE_BLENDTEXTUREALPHAPM: // Arg1 + Arg2 * (1 - Alpha)
{
- res.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), texture.a); res.a = MulHigh(res.a, arg2->a) << 4; res.a = AddSat(res.a, arg1->a);
+ res.w = SubSat(Short4(0x1000), texture.w); res.w = MulHigh(res.w, arg2->w) << 4; res.w = AddSat(res.w, arg1->w);
}
break;
case TextureStage::STAGE_PREMODULATE:
{
- res.a = arg1->a;
+ res.w = arg1->w;
}
break;
case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR:
@@ -1421,9 +1429,9 @@
case TextureStage::STAGE_ADDSIGNED2X:
case TextureStage::STAGE_SUBTRACT:
case TextureStage::STAGE_ADDSMOOTH:
- res.r = Max(res.r, Short4(0x0000, 0x0000, 0x0000, 0x0000));
- res.g = Max(res.g, Short4(0x0000, 0x0000, 0x0000, 0x0000));
- res.b = Max(res.b, Short4(0x0000, 0x0000, 0x0000, 0x0000));
+ res.x = Max(res.x, Short4(0x0000, 0x0000, 0x0000, 0x0000));
+ res.y = Max(res.y, Short4(0x0000, 0x0000, 0x0000, 0x0000));
+ res.z = Max(res.z, Short4(0x0000, 0x0000, 0x0000, 0x0000));
break;
default:
ASSERT(false);
@@ -1462,7 +1470,7 @@
case TextureStage::STAGE_ADDSIGNED2X:
case TextureStage::STAGE_SUBTRACT:
case TextureStage::STAGE_ADDSMOOTH:
- res.a = Max(res.a, Short4(0x0000, 0x0000, 0x0000, 0x0000));
+ res.w = Max(res.w, Short4(0x0000, 0x0000, 0x0000, 0x0000));
break;
default:
ASSERT(false);
@@ -1498,9 +1506,9 @@
case TextureStage::STAGE_BLENDTEXTUREALPHAPM:
case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR:
case TextureStage::STAGE_MODULATECOLOR_ADDALPHA:
- res.r = Min(res.r, Short4(0x1000, 0x1000, 0x1000, 0x1000));
- res.g = Min(res.g, Short4(0x1000, 0x1000, 0x1000, 0x1000));
- res.b = Min(res.b, Short4(0x1000, 0x1000, 0x1000, 0x1000));
+ res.x = Min(res.x, Short4(0x1000));
+ res.y = Min(res.y, Short4(0x1000));
+ res.z = Min(res.z, Short4(0x1000));
break;
default:
ASSERT(false);
@@ -1536,7 +1544,7 @@
case TextureStage::STAGE_BLENDTEXTUREALPHAPM:
case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR:
case TextureStage::STAGE_MODULATECOLOR_ADDALPHA:
- res.a = Min(res.a, Short4(0x1000, 0x1000, 0x1000, 0x1000));
+ res.w = Min(res.w, Short4(0x1000));
break;
default:
ASSERT(false);
@@ -1545,16 +1553,16 @@
switch(textureStage.destinationArgument)
{
case TextureStage::DESTINATION_CURRENT:
- current.r = res.r;
- current.g = res.g;
- current.b = res.b;
- current.a = res.a;
+ current.x = res.x;
+ current.y = res.y;
+ current.z = res.z;
+ current.w = res.w;
break;
case TextureStage::DESTINATION_TEMP:
- temp.r = res.r;
- temp.g = res.g;
- temp.b = res.b;
- temp.a = res.a;
+ temp.x = res.x;
+ temp.y = res.y;
+ temp.z = res.z;
+ temp.w = res.w;
break;
default:
ASSERT(false);
@@ -1623,7 +1631,7 @@
cMask[3] &= aMask3;
}
- Bool PixelRoutine::alphaTest(Registers &r, Int cMask[4], Color4i ¤t)
+ Bool PixelRoutine::alphaTest(Registers &r, Int cMask[4], Vector4i ¤t)
{
if(!state.alphaTestActive())
{
@@ -1634,7 +1642,7 @@
if(state.transparencyAntialiasing == Context::TRANSPARENCY_NONE)
{
- alphaTest(r, aMask, current.a);
+ alphaTest(r, aMask, current.w);
for(unsigned int q = 0; q < state.multiSample; q++)
{
@@ -1643,7 +1651,7 @@
}
else if(state.transparencyAntialiasing == Context::TRANSPARENCY_ALPHA_TO_COVERAGE)
{
- Float4 alpha = Float4(current.a) * Float4(1.0f / 0x1000);
+ Float4 alpha = Float4(current.w) * Float4(1.0f / 0x1000);
alphaToCoverage(r, cMask, alpha);
}
@@ -1659,7 +1667,7 @@
return pass != 0x0;
}
- Bool PixelRoutine::alphaTest(Registers &r, Int cMask[4], Color4f &c0)
+ Bool PixelRoutine::alphaTest(Registers &r, Int cMask[4], Vector4f &c0)
{
if(!state.alphaTestActive())
{
@@ -1670,7 +1678,7 @@
if(state.transparencyAntialiasing == Context::TRANSPARENCY_NONE)
{
- Short4 alpha = RoundShort4(c0.a * Float4(0x1000, 0x1000, 0x1000, 0x1000));
+ Short4 alpha = RoundShort4(c0.w * Float4(0x1000));
alphaTest(r, aMask, alpha);
@@ -1681,7 +1689,7 @@
}
else if(state.transparencyAntialiasing == Context::TRANSPARENCY_ALPHA_TO_COVERAGE)
{
- alphaToCoverage(r, cMask, c0.a);
+ alphaToCoverage(r, cMask, c0.w);
}
else ASSERT(false);
@@ -1695,7 +1703,7 @@
return pass != 0x0;
}
- void PixelRoutine::fogBlend(Registers &r, Color4i ¤t, Float4 &f, Float4 &z, Float4 &rhw)
+ void PixelRoutine::fogBlend(Registers &r, Vector4i ¤t, Float4 &f, Float4 &z, Float4 &rhw)
{
if(!state.fogActive)
{
@@ -1709,18 +1717,18 @@
UShort4 fog = convertFixed16(f, true);
- current.r = As<Short4>(MulHigh(As<UShort4>(current.r), fog));
- current.g = As<Short4>(MulHigh(As<UShort4>(current.g), fog));
- current.b = As<Short4>(MulHigh(As<UShort4>(current.b), fog));
+ current.x = As<Short4>(MulHigh(As<UShort4>(current.x), fog));
+ current.y = As<Short4>(MulHigh(As<UShort4>(current.y), fog));
+ current.z = As<Short4>(MulHigh(As<UShort4>(current.z), fog));
- UShort4 invFog = UShort4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - fog;
+ UShort4 invFog = UShort4(0xFFFFu) - fog;
- current.r += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(r.data + OFFSET(DrawData,fog.color4[0]))));
- current.g += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(r.data + OFFSET(DrawData,fog.color4[1]))));
- current.b += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(r.data + OFFSET(DrawData,fog.color4[2]))));
+ 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]))));
}
- void PixelRoutine::fogBlend(Registers &r, Color4f &c0, Float4 &fog, Float4 &z, Float4 &rhw)
+ void PixelRoutine::fogBlend(Registers &r, Vector4f &c0, Float4 &fog, Float4 &z, Float4 &rhw)
{
if(!state.fogActive)
{
@@ -1731,21 +1739,21 @@
{
pixelFog(r, fog, z, rhw);
- fog = Min(fog, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- fog = Max(fog, Float4(0.0f, 0.0f, 0.0f, 0.0f));
+ fog = Min(fog, Float4(1.0f));
+ fog = Max(fog, Float4(0.0f));
}
- c0.r -= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[0]));
- c0.g -= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[1]));
- c0.b -= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[2]));
+ 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.r *= fog;
- c0.g *= fog;
- c0.b *= fog;
+ c0.x *= fog;
+ c0.y *= fog;
+ c0.z *= fog;
- c0.r += *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[0]));
- c0.g += *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[1]));
- c0.b += *Pointer<Float4>(r.data + OFFSET(DrawData,fog.colorF[2]));
+ 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]));
}
void PixelRoutine::pixelFog(Registers &r, Float4 &visibility, Float4 &z, Float4 &rhw)
@@ -1762,7 +1770,7 @@
{
if(complementaryDepthBuffer)
{
- zw = Float4(1.0f, 1.0f, 1.0f, 1.0f) - z;
+ zw = Float4(1.0f) - z;
}
else
{
@@ -1781,12 +1789,12 @@
break;
case Context::FOG_EXP:
zw *= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.densityE));
- zw = exponential(zw, true);
+ zw = exponential2(zw, true);
break;
case Context::FOG_EXP2:
zw *= *Pointer<Float4>(r.data + OFFSET(DrawData,fog.densityE2));
zw *= zw;
- zw = exponential(zw, true);
+ zw = exponential2(zw, true);
zw = Rcp_pp(zw);
break;
default:
@@ -1794,16 +1802,16 @@
}
}
- void PixelRoutine::specularPixel(Color4i ¤t, Color4i &specular)
+ void PixelRoutine::specularPixel(Vector4i ¤t, Vector4i &specular)
{
if(!state.specularAdd)
{
return;
}
- current.r = AddSat(current.r, specular.r);
- current.g = AddSat(current.g, specular.g);
- current.b = AddSat(current.b, specular.b);
+ current.x = AddSat(current.x, specular.x);
+ current.y = AddSat(current.y, specular.y);
+ current.z = AddSat(current.z, specular.z);
}
void PixelRoutine::writeDepth(Registers &r, Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &zMask)
@@ -1815,11 +1823,11 @@
Float4 Z = z;
- if(pixelShader && pixelShader->depthOverride())
+ if(shader && shader->depthOverride())
{
if(complementaryDepthBuffer)
{
- Z = Float4(1, 1, 1, 1) - r.oDepth;
+ Z = Float4(1.0f) - r.oDepth;
}
else
{
@@ -2009,12 +2017,12 @@
}
}
- void PixelRoutine::sampleTexture(Registers &r, Color4i &c, int coordinates, int stage, bool project)
+ void PixelRoutine::sampleTexture(Registers &r, Vector4i &c, int coordinates, int stage, bool project)
{
- Float4 u = r.vx[2 + coordinates];
- Float4 v = r.vy[2 + coordinates];
- Float4 w = r.vz[2 + coordinates];
- Float4 q = r.vw[2 + coordinates];
+ Float4 u = r.vf[2 + coordinates].x;
+ Float4 v = r.vf[2 + coordinates].y;
+ Float4 w = r.vf[2 + coordinates].z;
+ Float4 q = r.vf[2 + coordinates].w;
if(perturbate)
{
@@ -2027,15 +2035,15 @@
sampleTexture(r, c, stage, u, v, w, q, project);
}
- void PixelRoutine::sampleTexture(Registers &r, Color4i &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project, bool bias, bool fixed12)
+ void PixelRoutine::sampleTexture(Registers &r, Vector4i &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project, bool bias, bool fixed12)
{
- Color4f dsx;
- Color4f dsy;
+ Vector4f dsx;
+ Vector4f dsy;
sampleTexture(r, c, stage, u, v, w, q, dsx, dsy, project, bias, fixed12, false);
}
- void PixelRoutine::sampleTexture(Registers &r, Color4i &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool project, bool bias, bool fixed12, bool gradients, bool lodProvided)
+ void PixelRoutine::sampleTexture(Registers &r, Vector4i &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project, bool bias, bool fixed12, bool gradients, bool lodProvided)
{
#if PERF_PROFILE
Long texTime = Ticks();
@@ -2063,7 +2071,31 @@
#endif
}
- void PixelRoutine::sampleTexture(Registers &r, Color4f &c, int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool project, bool bias, bool gradients, bool lodProvided)
+ void PixelRoutine::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)
+ {
+ 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);
+ }
+ else
+ {
+ Int index = As<Int>(Float(reg(r, sampler).x.x));
+
+ for(int i = 0; i < 16; i++)
+ {
+ if(shader->usesSampler(i))
+ {
+ If(index == i)
+ {
+ sampleTexture(r, 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
+ }
+ }
+ }
+ }
+ }
+
+ void PixelRoutine::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)
{
#if PERF_PROFILE
Long texTime = Ticks();
@@ -2091,7 +2123,7 @@
#endif
}
- void PixelRoutine::clampColor(Color4f oC[4])
+ void PixelRoutine::clampColor(Vector4f oC[4])
{
for(int index = 0; index < 4; index++)
{
@@ -2108,10 +2140,10 @@
case FORMAT_A8R8G8B8:
case FORMAT_X8R8G8B8:
case FORMAT_G16R16:
- oC[index].r = Max(oC[index].r, Float4(0.0f, 0.0f, 0.0f, 0.0f)); oC[index].r = Min(oC[index].r, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- oC[index].g = Max(oC[index].g, Float4(0.0f, 0.0f, 0.0f, 0.0f)); oC[index].g = Min(oC[index].g, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- oC[index].b = Max(oC[index].b, Float4(0.0f, 0.0f, 0.0f, 0.0f)); oC[index].b = Min(oC[index].b, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- oC[index].a = Max(oC[index].a, Float4(0.0f, 0.0f, 0.0f, 0.0f)); oC[index].a = Min(oC[index].a, Float4(1.0f, 1.0f, 1.0f, 1.0f));
+ oC[index].x = Max(oC[index].x, Float4(0.0f)); oC[index].x = Min(oC[index].x, Float4(1.0f));
+ oC[index].y = Max(oC[index].y, Float4(0.0f)); oC[index].y = Min(oC[index].y, Float4(1.0f));
+ oC[index].z = Max(oC[index].z, Float4(0.0f)); oC[index].z = Min(oC[index].z, Float4(1.0f));
+ oC[index].w = Max(oC[index].w, Float4(0.0f)); oC[index].w = Min(oC[index].w, Float4(1.0f));
break;
case FORMAT_R32F:
case FORMAT_G32R32F:
@@ -2123,14 +2155,14 @@
}
}
- void PixelRoutine::rasterOperation(Color4i ¤t, Registers &r, Float4 &fog, Pointer<Byte> &cBuffer, Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
+ void PixelRoutine::rasterOperation(Vector4i ¤t, Registers &r, Float4 &fog, Pointer<Byte> &cBuffer, Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
{
if(!state.colorWriteActive(0))
{
return;
}
- Color4f oC;
+ Vector4f oC;
switch(state.targetFormat[0])
{
@@ -2144,10 +2176,10 @@
}
else
{
- current.r <<= 4;
- current.g <<= 4;
- current.b <<= 4;
- current.a <<= 4;
+ current.x <<= 4;
+ current.y <<= 4;
+ current.z <<= 4;
+ current.w <<= 4;
}
fogBlend(r, current, fog, r.z[0], r.rhw);
@@ -2155,7 +2187,7 @@
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer + q * *Pointer<Int>(r.data + OFFSET(DrawData,colorSliceB[0]));
- Color4i color = current;
+ Vector4i color = current;
if(state.multiSampleMask & (1 << q))
{
@@ -2173,7 +2205,7 @@
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer + q * *Pointer<Int>(r.data + OFFSET(DrawData,colorSliceB[0]));
- Color4f color = oC;
+ Vector4f color = oC;
if(state.multiSampleMask & (1 << q))
{
@@ -2187,7 +2219,7 @@
}
}
- void PixelRoutine::rasterOperation(Color4f oC[4], Registers &r, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
+ void PixelRoutine::rasterOperation(Vector4f oC[4], Registers &r, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4])
{
for(int index = 0; index < 4; index++)
{
@@ -2198,9 +2230,9 @@
if(!postBlendSRGB && state.writeSRGB)
{
- oC[index].r = linearToSRGB(oC[index].r);
- oC[index].g = linearToSRGB(oC[index].g);
- oC[index].b = linearToSRGB(oC[index].b);
+ oC[index].x = linearToSRGB(oC[index].x);
+ oC[index].y = linearToSRGB(oC[index].y);
+ oC[index].z = linearToSRGB(oC[index].z);
}
if(index == 0)
@@ -2217,12 +2249,12 @@
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer[index] + q * *Pointer<Int>(r.data + OFFSET(DrawData,colorSliceB[index]));
- Color4i color;
+ Vector4i color;
- color.r = convertFixed16(oC[index].r, false);
- color.g = convertFixed16(oC[index].g, false);
- color.b = convertFixed16(oC[index].b, false);
- color.a = convertFixed16(oC[index].a, false);
+ color.x = convertFixed16(oC[index].x, false);
+ color.y = convertFixed16(oC[index].y, false);
+ color.z = convertFixed16(oC[index].z, false);
+ color.w = convertFixed16(oC[index].w, false);
if(state.multiSampleMask & (1 << q))
{
@@ -2237,7 +2269,7 @@
for(unsigned int q = 0; q < state.multiSample; q++)
{
Pointer<Byte> buffer = cBuffer[index] + q * *Pointer<Int>(r.data + OFFSET(DrawData,colorSliceB[index]));
- Color4f color = oC[index];
+ Vector4f color = oC[index];
if(state.multiSampleMask & (1 << q))
{
@@ -2252,7 +2284,7 @@
}
}
- void PixelRoutine::blendFactor(Registers &r, const Color4i &blendFactor, const Color4i ¤t, const Color4i &pixel, Context::BlendFactor blendFactorActive)
+ void PixelRoutine::blendFactor(Registers &r, const Vector4i &blendFactor, const Vector4i ¤t, const Vector4i &pixel, Context::BlendFactor blendFactorActive)
{
switch(blendFactorActive)
{
@@ -2263,77 +2295,77 @@
// Optimized
break;
case Context::BLEND_SOURCE:
- blendFactor.r = current.r;
- blendFactor.g = current.g;
- blendFactor.b = current.b;
+ blendFactor.x = current.x;
+ blendFactor.y = current.y;
+ blendFactor.z = current.z;
break;
case Context::BLEND_INVSOURCE:
- blendFactor.r = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.r;
- blendFactor.g = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.g;
- blendFactor.b = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.b;
+ blendFactor.x = Short4(0xFFFFu) - current.x;
+ blendFactor.y = Short4(0xFFFFu) - current.y;
+ blendFactor.z = Short4(0xFFFFu) - current.z;
break;
case Context::BLEND_DEST:
- blendFactor.r = pixel.r;
- blendFactor.g = pixel.g;
- blendFactor.b = pixel.b;
+ blendFactor.x = pixel.x;
+ blendFactor.y = pixel.y;
+ blendFactor.z = pixel.z;
break;
case Context::BLEND_INVDEST:
- blendFactor.r = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.r;
- blendFactor.g = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.g;
- blendFactor.b = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.b;
+ blendFactor.x = Short4(0xFFFFu) - pixel.x;
+ blendFactor.y = Short4(0xFFFFu) - pixel.y;
+ blendFactor.z = Short4(0xFFFFu) - pixel.z;
break;
case Context::BLEND_SOURCEALPHA:
- blendFactor.r = current.a;
- blendFactor.g = current.a;
- blendFactor.b = current.a;
+ blendFactor.x = current.w;
+ blendFactor.y = current.w;
+ blendFactor.z = current.w;
break;
case Context::BLEND_INVSOURCEALPHA:
- blendFactor.r = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.a;
- blendFactor.g = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.a;
- blendFactor.b = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.a;
+ blendFactor.x = Short4(0xFFFFu) - current.w;
+ blendFactor.y = Short4(0xFFFFu) - current.w;
+ blendFactor.z = Short4(0xFFFFu) - current.w;
break;
case Context::BLEND_DESTALPHA:
- blendFactor.r = pixel.a;
- blendFactor.g = pixel.a;
- blendFactor.b = pixel.a;
+ blendFactor.x = pixel.w;
+ blendFactor.y = pixel.w;
+ blendFactor.z = pixel.w;
break;
case Context::BLEND_INVDESTALPHA:
- blendFactor.r = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.a;
- blendFactor.g = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.a;
- blendFactor.b = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.a;
+ blendFactor.x = Short4(0xFFFFu) - pixel.w;
+ blendFactor.y = Short4(0xFFFFu) - pixel.w;
+ blendFactor.z = Short4(0xFFFFu) - pixel.w;
break;
case Context::BLEND_SRCALPHASAT:
- blendFactor.r = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.a;
- blendFactor.r = Min(As<UShort4>(blendFactor.r), As<UShort4>(current.a));
- blendFactor.g = blendFactor.r;
- blendFactor.b = blendFactor.r;
+ blendFactor.x = Short4(0xFFFFu) - pixel.w;
+ blendFactor.x = Min(As<UShort4>(blendFactor.x), As<UShort4>(current.w));
+ blendFactor.y = blendFactor.x;
+ blendFactor.z = blendFactor.x;
break;
case Context::BLEND_CONSTANT:
- blendFactor.r = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[0]));
- blendFactor.g = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[1]));
- blendFactor.b = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[2]));
+ 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]));
break;
case Context::BLEND_INVCONSTANT:
- blendFactor.r = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[0]));
- blendFactor.g = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[1]));
- blendFactor.b = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[2]));
+ 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]));
break;
case Context::BLEND_CONSTANTALPHA:
- blendFactor.r = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
- blendFactor.g = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
- blendFactor.b = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
+ 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]));
break;
case Context::BLEND_INVCONSTANTALPHA:
- blendFactor.r = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
- blendFactor.g = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
- blendFactor.b = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
+ 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]));
break;
default:
ASSERT(false);
}
}
- void PixelRoutine::blendFactorAlpha(Registers &r, const Color4i &blendFactor, const Color4i ¤t, const Color4i &pixel, Context::BlendFactor blendFactorAlphaActive)
+ void PixelRoutine::blendFactorAlpha(Registers &r, const Vector4i &blendFactor, const Vector4i ¤t, const Vector4i &pixel, Context::BlendFactor blendFactorAlphaActive)
{
switch(blendFactorAlphaActive)
{
@@ -2344,46 +2376,46 @@
// Optimized
break;
case Context::BLEND_SOURCE:
- blendFactor.a = current.a;
+ blendFactor.w = current.w;
break;
case Context::BLEND_INVSOURCE:
- blendFactor.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.a;
+ blendFactor.w = Short4(0xFFFFu) - current.w;
break;
case Context::BLEND_DEST:
- blendFactor.a = pixel.a;
+ blendFactor.w = pixel.w;
break;
case Context::BLEND_INVDEST:
- blendFactor.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.a;
+ blendFactor.w = Short4(0xFFFFu) - pixel.w;
break;
case Context::BLEND_SOURCEALPHA:
- blendFactor.a = current.a;
+ blendFactor.w = current.w;
break;
case Context::BLEND_INVSOURCEALPHA:
- blendFactor.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - current.a;
+ blendFactor.w = Short4(0xFFFFu) - current.w;
break;
case Context::BLEND_DESTALPHA:
- blendFactor.a = pixel.a;
+ blendFactor.w = pixel.w;
break;
case Context::BLEND_INVDESTALPHA:
- blendFactor.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF) - pixel.a;
+ blendFactor.w = Short4(0xFFFFu) - pixel.w;
break;
case Context::BLEND_SRCALPHASAT:
- blendFactor.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF);
+ blendFactor.w = Short4(0xFFFFu);
break;
case Context::BLEND_CONSTANT:
case Context::BLEND_CONSTANTALPHA:
- blendFactor.a = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
+ blendFactor.w = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.blendConstant4W[3]));
break;
case Context::BLEND_INVCONSTANT:
case Context::BLEND_INVCONSTANTALPHA:
- blendFactor.a = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
+ blendFactor.w = *Pointer<Short4>(r.data + OFFSET(DrawData,factor.invBlendConstant4W[3]));
break;
default:
ASSERT(false);
}
}
- void PixelRoutine::alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Color4i ¤t, Int &x)
+ void PixelRoutine::alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4i ¤t, Int &x)
{
if(!state.alphaBlendActive)
{
@@ -2392,7 +2424,7 @@
Pointer<Byte> buffer;
- Color4i pixel;
+ Vector4i pixel;
Short4 c01;
Short4 c23;
@@ -2404,74 +2436,74 @@
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
- pixel.b = c01;
- pixel.g = c01;
- pixel.b = UnpackLow(As<Byte8>(pixel.b), As<Byte8>(c23));
- pixel.g = UnpackHigh(As<Byte8>(pixel.g), As<Byte8>(c23));
- pixel.r = pixel.b;
- pixel.b = UnpackLow(As<Byte8>(pixel.b), As<Byte8>(pixel.g));
- pixel.r = UnpackHigh(As<Byte8>(pixel.r), As<Byte8>(pixel.g));
- pixel.g = pixel.b;
- pixel.a = pixel.r;
- pixel.r = UnpackLow(As<Byte8>(pixel.r), As<Byte8>(pixel.r));
- pixel.g = UnpackHigh(As<Byte8>(pixel.g), As<Byte8>(pixel.g));
- pixel.b = UnpackLow(As<Byte8>(pixel.b), As<Byte8>(pixel.b));
- pixel.a = UnpackHigh(As<Byte8>(pixel.a), As<Byte8>(pixel.a));
+ pixel.z = c01;
+ pixel.y = c01;
+ pixel.z = UnpackLow(As<Byte8>(pixel.z), As<Byte8>(c23));
+ pixel.y = UnpackHigh(As<Byte8>(pixel.y), As<Byte8>(c23));
+ pixel.x = pixel.z;
+ pixel.z = UnpackLow(As<Byte8>(pixel.z), As<Byte8>(pixel.y));
+ pixel.x = UnpackHigh(As<Byte8>(pixel.x), As<Byte8>(pixel.y));
+ pixel.y = pixel.z;
+ pixel.w = pixel.x;
+ pixel.x = UnpackLow(As<Byte8>(pixel.x), As<Byte8>(pixel.x));
+ pixel.y = UnpackHigh(As<Byte8>(pixel.y), As<Byte8>(pixel.y));
+ pixel.z = UnpackLow(As<Byte8>(pixel.z), As<Byte8>(pixel.z));
+ pixel.w = UnpackHigh(As<Byte8>(pixel.w), As<Byte8>(pixel.w));
break;
case FORMAT_X8R8G8B8:
buffer = cBuffer + 4 * x;
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
- pixel.b = c01;
- pixel.g = c01;
- pixel.b = UnpackLow(As<Byte8>(pixel.b), As<Byte8>(c23));
- pixel.g = UnpackHigh(As<Byte8>(pixel.g), As<Byte8>(c23));
- pixel.r = pixel.b;
- pixel.b = UnpackLow(As<Byte8>(pixel.b), As<Byte8>(pixel.g));
- pixel.r = UnpackHigh(As<Byte8>(pixel.r), As<Byte8>(pixel.g));
- pixel.g = pixel.b;
- pixel.r = UnpackLow(As<Byte8>(pixel.r), As<Byte8>(pixel.r));
- pixel.g = UnpackHigh(As<Byte8>(pixel.g), As<Byte8>(pixel.g));
- pixel.b = UnpackLow(As<Byte8>(pixel.b), As<Byte8>(pixel.b));
- pixel.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF);
+ pixel.z = c01;
+ pixel.y = c01;
+ pixel.z = UnpackLow(As<Byte8>(pixel.z), As<Byte8>(c23));
+ pixel.y = UnpackHigh(As<Byte8>(pixel.y), As<Byte8>(c23));
+ pixel.x = pixel.z;
+ pixel.z = UnpackLow(As<Byte8>(pixel.z), As<Byte8>(pixel.y));
+ pixel.x = UnpackHigh(As<Byte8>(pixel.x), As<Byte8>(pixel.y));
+ pixel.y = pixel.z;
+ pixel.x = UnpackLow(As<Byte8>(pixel.x), As<Byte8>(pixel.x));
+ pixel.y = UnpackHigh(As<Byte8>(pixel.y), As<Byte8>(pixel.y));
+ pixel.z = UnpackLow(As<Byte8>(pixel.z), As<Byte8>(pixel.z));
+ pixel.w = Short4(0xFFFFu);
break;
case FORMAT_A8G8R8B8Q:
UNIMPLEMENTED();
- // pixel.b = UnpackLow(As<Byte8>(pixel.b), *Pointer<Byte8>(cBuffer + 8 * x + 0));
- // pixel.r = UnpackHigh(As<Byte8>(pixel.r), *Pointer<Byte8>(cBuffer + 8 * x + 0));
- // pixel.g = UnpackLow(As<Byte8>(pixel.g), *Pointer<Byte8>(cBuffer + 8 * x + 8));
- // pixel.a = UnpackHigh(As<Byte8>(pixel.a), *Pointer<Byte8>(cBuffer + 8 * x + 8));
+ // pixel.z = UnpackLow(As<Byte8>(pixel.z), *Pointer<Byte8>(cBuffer + 8 * x + 0));
+ // pixel.x = UnpackHigh(As<Byte8>(pixel.x), *Pointer<Byte8>(cBuffer + 8 * x + 0));
+ // pixel.y = UnpackLow(As<Byte8>(pixel.y), *Pointer<Byte8>(cBuffer + 8 * x + 8));
+ // pixel.w = UnpackHigh(As<Byte8>(pixel.w), *Pointer<Byte8>(cBuffer + 8 * x + 8));
break;
case FORMAT_X8G8R8B8Q:
UNIMPLEMENTED();
- // pixel.b = UnpackLow(As<Byte8>(pixel.b), *Pointer<Byte8>(cBuffer + 8 * x + 0));
- // pixel.r = UnpackHigh(As<Byte8>(pixel.r), *Pointer<Byte8>(cBuffer + 8 * x + 0));
- // pixel.g = UnpackLow(As<Byte8>(pixel.g), *Pointer<Byte8>(cBuffer + 8 * x + 8));
- // pixel.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF);
+ // pixel.z = UnpackLow(As<Byte8>(pixel.z), *Pointer<Byte8>(cBuffer + 8 * x + 0));
+ // pixel.x = UnpackHigh(As<Byte8>(pixel.x), *Pointer<Byte8>(cBuffer + 8 * x + 0));
+ // pixel.y = UnpackLow(As<Byte8>(pixel.y), *Pointer<Byte8>(cBuffer + 8 * x + 8));
+ // pixel.w = Short4(0xFFFFu);
break;
case FORMAT_A16B16G16R16:
buffer = cBuffer;
- pixel.r = *Pointer<Short4>(buffer + 8 * x);
- pixel.g = *Pointer<Short4>(buffer + 8 * x + 8);
+ pixel.x = *Pointer<Short4>(buffer + 8 * x);
+ pixel.y = *Pointer<Short4>(buffer + 8 * x + 8);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
- pixel.b = *Pointer<Short4>(buffer + 8 * x);
- pixel.a = *Pointer<Short4>(buffer + 8 * x + 8);
- transpose4x4(pixel.r, pixel.g, pixel.b, pixel.a);
+ pixel.z = *Pointer<Short4>(buffer + 8 * x);
+ pixel.w = *Pointer<Short4>(buffer + 8 * x + 8);
+ transpose4x4(pixel.x, pixel.y, pixel.z, pixel.w);
break;
case FORMAT_G16R16:
buffer = cBuffer;
- pixel.r = *Pointer<Short4>(buffer + 4 * x);
+ pixel.x = *Pointer<Short4>(buffer + 4 * x);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
- pixel.g = *Pointer<Short4>(buffer + 4 * x);
- pixel.b = pixel.r;
- pixel.r = As<Short4>(UnpackLow(pixel.r, pixel.g));
- pixel.b = As<Short4>(UnpackHigh(pixel.b, pixel.g));
- pixel.g = pixel.b;
- pixel.r = As<Short4>(UnpackLow(pixel.r, pixel.b));
- pixel.g = As<Short4>(UnpackHigh(pixel.g, pixel.b));
- pixel.b = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF);
- pixel.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF);
+ pixel.y = *Pointer<Short4>(buffer + 4 * x);
+ pixel.z = pixel.x;
+ pixel.x = As<Short4>(UnpackLow(pixel.x, pixel.y));
+ pixel.z = As<Short4>(UnpackHigh(pixel.z, pixel.y));
+ pixel.y = pixel.z;
+ pixel.x = As<Short4>(UnpackLow(pixel.x, pixel.z));
+ pixel.y = As<Short4>(UnpackHigh(pixel.y, pixel.z));
+ pixel.z = Short4(0xFFFFu);
+ pixel.w = Short4(0xFFFFu);
break;
default:
ASSERT(false);
@@ -2483,65 +2515,65 @@
}
// Final Color = ObjectColor * SourceBlendFactor + PixelColor * DestinationBlendFactor
- Color4i sourceFactor;
- Color4i destFactor;
+ Vector4i sourceFactor;
+ Vector4i destFactor;
blendFactor(r, sourceFactor, current, pixel, (Context::BlendFactor)state.sourceBlendFactor);
blendFactor(r, destFactor, current, pixel, (Context::BlendFactor)state.destBlendFactor);
if(state.sourceBlendFactor != Context::BLEND_ONE && state.sourceBlendFactor != Context::BLEND_ZERO)
{
- current.r = MulHigh(As<UShort4>(current.r), As<UShort4>(sourceFactor.r));
- current.g = MulHigh(As<UShort4>(current.g), As<UShort4>(sourceFactor.g));
- current.b = MulHigh(As<UShort4>(current.b), As<UShort4>(sourceFactor.b));
+ current.x = MulHigh(As<UShort4>(current.x), As<UShort4>(sourceFactor.x));
+ current.y = MulHigh(As<UShort4>(current.y), As<UShort4>(sourceFactor.y));
+ current.z = MulHigh(As<UShort4>(current.z), As<UShort4>(sourceFactor.z));
}
if(state.destBlendFactor != Context::BLEND_ONE && state.destBlendFactor != Context::BLEND_ZERO)
{
- pixel.r = MulHigh(As<UShort4>(pixel.r), As<UShort4>(destFactor.r));
- pixel.g = MulHigh(As<UShort4>(pixel.g), As<UShort4>(destFactor.g));
- pixel.b = MulHigh(As<UShort4>(pixel.b), As<UShort4>(destFactor.b));
+ pixel.x = MulHigh(As<UShort4>(pixel.x), As<UShort4>(destFactor.x));
+ pixel.y = MulHigh(As<UShort4>(pixel.y), As<UShort4>(destFactor.y));
+ pixel.z = MulHigh(As<UShort4>(pixel.z), As<UShort4>(destFactor.z));
}
switch(state.blendOperation)
{
case Context::BLENDOP_ADD:
- current.r = AddSat(As<UShort4>(current.r), As<UShort4>(pixel.r));
- current.g = AddSat(As<UShort4>(current.g), As<UShort4>(pixel.g));
- current.b = AddSat(As<UShort4>(current.b), As<UShort4>(pixel.b));
+ current.x = AddSat(As<UShort4>(current.x), As<UShort4>(pixel.x));
+ current.y = AddSat(As<UShort4>(current.y), As<UShort4>(pixel.y));
+ current.z = AddSat(As<UShort4>(current.z), As<UShort4>(pixel.z));
break;
case Context::BLENDOP_SUB:
- current.r = SubSat(As<UShort4>(current.r), As<UShort4>(pixel.r));
- current.g = SubSat(As<UShort4>(current.g), As<UShort4>(pixel.g));
- current.b = SubSat(As<UShort4>(current.b), As<UShort4>(pixel.b));
+ current.x = SubSat(As<UShort4>(current.x), As<UShort4>(pixel.x));
+ current.y = SubSat(As<UShort4>(current.y), As<UShort4>(pixel.y));
+ current.z = SubSat(As<UShort4>(current.z), As<UShort4>(pixel.z));
break;
case Context::BLENDOP_INVSUB:
- current.r = SubSat(As<UShort4>(pixel.r), As<UShort4>(current.r));
- current.g = SubSat(As<UShort4>(pixel.g), As<UShort4>(current.g));
- current.b = SubSat(As<UShort4>(pixel.b), As<UShort4>(current.b));
+ current.x = SubSat(As<UShort4>(pixel.x), As<UShort4>(current.x));
+ current.y = SubSat(As<UShort4>(pixel.y), As<UShort4>(current.y));
+ current.z = SubSat(As<UShort4>(pixel.z), As<UShort4>(current.z));
break;
case Context::BLENDOP_MIN:
- current.r = Min(As<UShort4>(current.r), As<UShort4>(pixel.r));
- current.g = Min(As<UShort4>(current.g), As<UShort4>(pixel.g));
- current.b = Min(As<UShort4>(current.b), As<UShort4>(pixel.b));
+ current.x = Min(As<UShort4>(current.x), As<UShort4>(pixel.x));
+ current.y = Min(As<UShort4>(current.y), As<UShort4>(pixel.y));
+ current.z = Min(As<UShort4>(current.z), As<UShort4>(pixel.z));
break;
case Context::BLENDOP_MAX:
- current.r = Max(As<UShort4>(current.r), As<UShort4>(pixel.r));
- current.g = Max(As<UShort4>(current.g), As<UShort4>(pixel.g));
- current.b = Max(As<UShort4>(current.b), As<UShort4>(pixel.b));
+ current.x = Max(As<UShort4>(current.x), As<UShort4>(pixel.x));
+ current.y = Max(As<UShort4>(current.y), As<UShort4>(pixel.y));
+ current.z = Max(As<UShort4>(current.z), As<UShort4>(pixel.z));
break;
case Context::BLENDOP_SOURCE:
// No operation
break;
case Context::BLENDOP_DEST:
- current.r = pixel.r;
- current.g = pixel.g;
- current.b = pixel.b;
+ current.x = pixel.x;
+ current.y = pixel.y;
+ current.z = pixel.z;
break;
case Context::BLENDOP_NULL:
- current.r = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- current.g = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- current.b = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ current.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ current.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ current.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
break;
default:
ASSERT(false);
@@ -2552,46 +2584,46 @@
if(state.sourceBlendFactorAlpha != Context::BLEND_ONE && state.sourceBlendFactorAlpha != Context::BLEND_ZERO)
{
- current.a = MulHigh(As<UShort4>(current.a), As<UShort4>(sourceFactor.a));
+ current.w = MulHigh(As<UShort4>(current.w), As<UShort4>(sourceFactor.w));
}
if(state.destBlendFactorAlpha != Context::BLEND_ONE && state.destBlendFactorAlpha != Context::BLEND_ZERO)
{
- pixel.a = MulHigh(As<UShort4>(pixel.a), As<UShort4>(destFactor.a));
+ pixel.w = MulHigh(As<UShort4>(pixel.w), As<UShort4>(destFactor.w));
}
switch(state.blendOperationAlpha)
{
case Context::BLENDOP_ADD:
- current.a = AddSat(As<UShort4>(current.a), As<UShort4>(pixel.a));
+ current.w = AddSat(As<UShort4>(current.w), As<UShort4>(pixel.w));
break;
case Context::BLENDOP_SUB:
- current.a = SubSat(As<UShort4>(current.a), As<UShort4>(pixel.a));
+ current.w = SubSat(As<UShort4>(current.w), As<UShort4>(pixel.w));
break;
case Context::BLENDOP_INVSUB:
- current.a = SubSat(As<UShort4>(pixel.a), As<UShort4>(current.a));
+ current.w = SubSat(As<UShort4>(pixel.w), As<UShort4>(current.w));
break;
case Context::BLENDOP_MIN:
- current.a = Min(As<UShort4>(current.a), As<UShort4>(pixel.a));
+ current.w = Min(As<UShort4>(current.w), As<UShort4>(pixel.w));
break;
case Context::BLENDOP_MAX:
- current.a = Max(As<UShort4>(current.a), As<UShort4>(pixel.a));
+ current.w = Max(As<UShort4>(current.w), As<UShort4>(pixel.w));
break;
case Context::BLENDOP_SOURCE:
// No operation
break;
case Context::BLENDOP_DEST:
- current.a = pixel.a;
+ current.w = pixel.w;
break;
case Context::BLENDOP_NULL:
- current.a = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ current.w = Short4(0x0000, 0x0000, 0x0000, 0x0000);
break;
default:
ASSERT(false);
}
}
- void PixelRoutine::writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Color4i ¤t, Int &sMask, Int &zMask, Int &cMask)
+ void PixelRoutine::writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Vector4i ¤t, Int &sMask, Int &zMask, Int &cMask)
{
if(!state.colorWriteActive(index))
{
@@ -2612,10 +2644,10 @@
case FORMAT_X8R8G8B8:
case FORMAT_A8R8G8B8:
{
- current.r = current.r - As<Short4>(As<UShort4>(current.r) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
- current.g = current.g - As<Short4>(As<UShort4>(current.g) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
- current.b = current.b - As<Short4>(As<UShort4>(current.b) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
- current.a = current.a - As<Short4>(As<UShort4>(current.a) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
+ current.x = current.x - As<Short4>(As<UShort4>(current.x) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
+ current.y = current.y - As<Short4>(As<UShort4>(current.y) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
+ current.z = current.z - As<Short4>(As<UShort4>(current.z) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
+ current.w = current.w - As<Short4>(As<UShort4>(current.w) >> 8) + Short4(0x0080, 0x0080, 0x0080, 0x0080);
}
break;
}
@@ -2629,78 +2661,78 @@
{
case FORMAT_X8G8R8B8Q:
UNIMPLEMENTED();
- // current.r = As<Short4>(As<UShort4>(current.r) >> 8);
- // current.g = As<Short4>(As<UShort4>(current.g) >> 8);
- // current.b = As<Short4>(As<UShort4>(current.b) >> 8);
+ // current.x = As<Short4>(As<UShort4>(current.x) >> 8);
+ // current.y = As<Short4>(As<UShort4>(current.y) >> 8);
+ // current.z = As<Short4>(As<UShort4>(current.z) >> 8);
- // current.b = As<Short4>(Pack(As<UShort4>(current.b), As<UShort4>(current.r)));
- // current.g = As<Short4>(Pack(As<UShort4>(current.g), As<UShort4>(current.g)));
+ // current.z = As<Short4>(Pack(As<UShort4>(current.z), As<UShort4>(current.x)));
+ // current.y = As<Short4>(Pack(As<UShort4>(current.y), As<UShort4>(current.y)));
break;
case FORMAT_A8G8R8B8Q:
UNIMPLEMENTED();
- // current.r = As<Short4>(As<UShort4>(current.r) >> 8);
- // current.g = As<Short4>(As<UShort4>(current.g) >> 8);
- // current.b = As<Short4>(As<UShort4>(current.b) >> 8);
- // current.a = As<Short4>(As<UShort4>(current.a) >> 8);
+ // current.x = As<Short4>(As<UShort4>(current.x) >> 8);
+ // current.y = As<Short4>(As<UShort4>(current.y) >> 8);
+ // current.z = As<Short4>(As<UShort4>(current.z) >> 8);
+ // current.w = As<Short4>(As<UShort4>(current.w) >> 8);
- // current.b = As<Short4>(Pack(As<UShort4>(current.b), As<UShort4>(current.r)));
- // current.g = As<Short4>(Pack(As<UShort4>(current.g), As<UShort4>(current.a)));
+ // current.z = As<Short4>(Pack(As<UShort4>(current.z), As<UShort4>(current.x)));
+ // current.y = As<Short4>(Pack(As<UShort4>(current.y), As<UShort4>(current.w)));
break;
case FORMAT_X8R8G8B8:
case FORMAT_A8R8G8B8:
if(state.targetFormat[index] == FORMAT_X8R8G8B8 || rgbaWriteMask == 0x7)
{
- current.r = As<Short4>(As<UShort4>(current.r) >> 8);
- current.g = As<Short4>(As<UShort4>(current.g) >> 8);
- current.b = As<Short4>(As<UShort4>(current.b) >> 8);
+ current.x = As<Short4>(As<UShort4>(current.x) >> 8);
+ current.y = As<Short4>(As<UShort4>(current.y) >> 8);
+ current.z = As<Short4>(As<UShort4>(current.z) >> 8);
- current.b = As<Short4>(Pack(As<UShort4>(current.b), As<UShort4>(current.r)));
- current.g = As<Short4>(Pack(As<UShort4>(current.g), As<UShort4>(current.g)));
+ current.z = As<Short4>(Pack(As<UShort4>(current.z), As<UShort4>(current.x)));
+ current.y = As<Short4>(Pack(As<UShort4>(current.y), As<UShort4>(current.y)));
- current.r = current.b;
- current.b = UnpackLow(As<Byte8>(current.b), As<Byte8>(current.g));
- current.r = UnpackHigh(As<Byte8>(current.r), As<Byte8>(current.g));
- current.g = current.b;
- current.b = As<Short4>(UnpackLow(current.b, current.r));
- current.g = As<Short4>(UnpackHigh(current.g, current.r));
+ current.x = current.z;
+ current.z = UnpackLow(As<Byte8>(current.z), As<Byte8>(current.y));
+ current.x = UnpackHigh(As<Byte8>(current.x), As<Byte8>(current.y));
+ current.y = current.z;
+ current.z = As<Short4>(UnpackLow(current.z, current.x));
+ current.y = As<Short4>(UnpackHigh(current.y, current.x));
}
else
{
- current.r = As<Short4>(As<UShort4>(current.r) >> 8);
- current.g = As<Short4>(As<UShort4>(current.g) >> 8);
- current.b = As<Short4>(As<UShort4>(current.b) >> 8);
- current.a = As<Short4>(As<UShort4>(current.a) >> 8);
+ current.x = As<Short4>(As<UShort4>(current.x) >> 8);
+ current.y = As<Short4>(As<UShort4>(current.y) >> 8);
+ current.z = As<Short4>(As<UShort4>(current.z) >> 8);
+ current.w = As<Short4>(As<UShort4>(current.w) >> 8);
- current.b = As<Short4>(Pack(As<UShort4>(current.b), As<UShort4>(current.r)));
- current.g = As<Short4>(Pack(As<UShort4>(current.g), As<UShort4>(current.a)));
+ current.z = As<Short4>(Pack(As<UShort4>(current.z), As<UShort4>(current.x)));
+ current.y = As<Short4>(Pack(As<UShort4>(current.y), As<UShort4>(current.w)));
- current.r = current.b;
- current.b = UnpackLow(As<Byte8>(current.b), As<Byte8>(current.g));
- current.r = UnpackHigh(As<Byte8>(current.r), As<Byte8>(current.g));
- current.g = current.b;
- current.b = As<Short4>(UnpackLow(current.b, current.r));
- current.g = As<Short4>(UnpackHigh(current.g, current.r));
+ current.x = current.z;
+ current.z = UnpackLow(As<Byte8>(current.z), As<Byte8>(current.y));
+ current.x = UnpackHigh(As<Byte8>(current.x), As<Byte8>(current.y));
+ current.y = current.z;
+ current.z = As<Short4>(UnpackLow(current.z, current.x));
+ current.y = As<Short4>(UnpackHigh(current.y, current.x));
}
break;
case FORMAT_G16R16:
- current.b = current.r;
- current.r = As<Short4>(UnpackLow(current.r, current.g));
- current.b = As<Short4>(UnpackHigh(current.b, current.g));
- current.g = current.b;
+ current.z = current.x;
+ current.x = As<Short4>(UnpackLow(current.x, current.y));
+ current.z = As<Short4>(UnpackHigh(current.z, current.y));
+ current.y = current.z;
break;
case FORMAT_A16B16G16R16:
- transpose4x4(current.r, current.g, current.b, current.a);
+ transpose4x4(current.x, current.y, current.z, current.w);
break;
case FORMAT_R32F:
case FORMAT_G32R32F:
case FORMAT_A32B32G32R32F:
{
- Color4f oC;
+ Vector4f oC;
- oC.r = convertUnsigned16(UShort4(current.r));
- oC.g = convertUnsigned16(UShort4(current.g));
- oC.b = convertUnsigned16(UShort4(current.b));
- oC.a = convertUnsigned16(UShort4(current.a));
+ oC.x = convertUnsigned16(UShort4(current.x));
+ oC.y = convertUnsigned16(UShort4(current.y));
+ oC.z = convertUnsigned16(UShort4(current.z));
+ oC.w = convertUnsigned16(UShort4(current.w));
writeColor(r, index, cBuffer, x, oC, sMask, zMask, cMask);
}
@@ -2709,8 +2741,8 @@
ASSERT(false);
}
- Short4 c01 = current.b;
- Short4 c23 = current.g;
+ Short4 c01 = current.z;
+ Short4 c23 = current.y;
Int xMask; // Combination of all masks
@@ -2816,15 +2848,15 @@
if((rgbaWriteMask & 0x00000003) != 0x00000003)
{
Short4 masked = value;
- current.r &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
+ current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW01Q[rgbaWriteMask & 0x3][0]));
- current.r |= masked;
+ current.x |= masked;
}
- current.r &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD01Q) + xMask * 8);
+ current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD01Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD01Q) + xMask * 8);
- current.r |= value;
- *Pointer<Short4>(buffer) = current.r;
+ current.x |= value;
+ *Pointer<Short4>(buffer) = current.x;
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
@@ -2833,15 +2865,15 @@
if((rgbaWriteMask & 0x00000003) != 0x00000003)
{
Short4 masked = value;
- current.g &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
+ current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW01Q[rgbaWriteMask & 0x3][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW01Q[rgbaWriteMask & 0x3][0]));
- current.g |= masked;
+ current.y |= masked;
}
- current.g &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD23Q) + xMask * 8);
+ current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskD23Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskD23Q) + xMask * 8);
- current.g |= value;
- *Pointer<Short4>(buffer) = current.g;
+ current.y |= value;
+ *Pointer<Short4>(buffer) = current.y;
break;
case FORMAT_A16B16G16R16:
buffer = cBuffer + 8 * x;
@@ -2852,15 +2884,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
- current.r &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
+ current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
- current.r |= masked;
+ current.x |= masked;
}
- current.r &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ0Q) + xMask * 8);
+ current.x &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ0Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ0Q) + xMask * 8);
- current.r |= value;
- *Pointer<Short4>(buffer) = current.r;
+ current.x |= value;
+ *Pointer<Short4>(buffer) = current.x;
}
{
@@ -2869,15 +2901,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
- current.g &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
+ current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
- current.g |= masked;
+ current.y |= masked;
}
- current.g &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ1Q) + xMask * 8);
+ current.y &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ1Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ1Q) + xMask * 8);
- current.g |= value;
- *Pointer<Short4>(buffer + 8) = current.g;
+ current.y |= value;
+ *Pointer<Short4>(buffer + 8) = current.y;
}
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
@@ -2888,15 +2920,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
- current.b &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
+ current.z &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
- current.b |= masked;
+ current.z |= masked;
}
- current.b &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ2Q) + xMask * 8);
+ current.z &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ2Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ2Q) + xMask * 8);
- current.b |= value;
- *Pointer<Short4>(buffer) = current.b;
+ current.z |= value;
+ *Pointer<Short4>(buffer) = current.z;
}
{
@@ -2905,15 +2937,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Short4 masked = value;
- current.a &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
+ current.w &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskW4Q[rgbaWriteMask][0]));
masked &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskW4Q[rgbaWriteMask][0]));
- current.a |= masked;
+ current.w |= masked;
}
- current.a &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ3Q) + xMask * 8);
+ current.w &= *Pointer<Short4>(r.constants + OFFSET(Constants,maskQ3Q) + xMask * 8);
value &= *Pointer<Short4>(r.constants + OFFSET(Constants,invMaskQ3Q) + xMask * 8);
- current.a |= value;
- *Pointer<Short4>(buffer + 8) = current.a;
+ current.w |= value;
+ *Pointer<Short4>(buffer + 8) = current.w;
}
break;
default:
@@ -2921,7 +2953,7 @@
}
}
- void PixelRoutine::blendFactor(Registers &r, const Color4f &blendFactor, const Color4f &oC, const Color4f &pixel, Context::BlendFactor blendFactorActive)
+ void PixelRoutine::blendFactor(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, Context::BlendFactor blendFactorActive)
{
switch(blendFactorActive)
{
@@ -2932,67 +2964,67 @@
// Optimized
break;
case Context::BLEND_SOURCE:
- blendFactor.r = oC.r;
- blendFactor.g = oC.g;
- blendFactor.b = oC.b;
+ blendFactor.x = oC.x;
+ blendFactor.y = oC.y;
+ blendFactor.z = oC.z;
break;
case Context::BLEND_INVSOURCE:
- blendFactor.r = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.r;
- blendFactor.g = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.g;
- blendFactor.b = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.b;
+ blendFactor.x = Float4(1.0f) - oC.x;
+ blendFactor.y = Float4(1.0f) - oC.y;
+ blendFactor.z = Float4(1.0f) - oC.z;
break;
case Context::BLEND_DEST:
- blendFactor.r = pixel.r;
- blendFactor.g = pixel.g;
- blendFactor.b = pixel.b;
+ blendFactor.x = pixel.x;
+ blendFactor.y = pixel.y;
+ blendFactor.z = pixel.z;
break;
case Context::BLEND_INVDEST:
- blendFactor.r = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.r;
- blendFactor.g = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.g;
- blendFactor.b = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.b;
+ blendFactor.x = Float4(1.0f) - pixel.x;
+ blendFactor.y = Float4(1.0f) - pixel.y;
+ blendFactor.z = Float4(1.0f) - pixel.z;
break;
case Context::BLEND_SOURCEALPHA:
- blendFactor.r = oC.a;
- blendFactor.g = oC.a;
- blendFactor.b = oC.a;
+ blendFactor.x = oC.w;
+ blendFactor.y = oC.w;
+ blendFactor.z = oC.w;
break;
case Context::BLEND_INVSOURCEALPHA:
- blendFactor.r = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.a;
- blendFactor.g = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.a;
- blendFactor.b = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.a;
+ blendFactor.x = Float4(1.0f) - oC.w;
+ blendFactor.y = Float4(1.0f) - oC.w;
+ blendFactor.z = Float4(1.0f) - oC.w;
break;
case Context::BLEND_DESTALPHA:
- blendFactor.r = pixel.a;
- blendFactor.g = pixel.a;
- blendFactor.b = pixel.a;
+ blendFactor.x = pixel.w;
+ blendFactor.y = pixel.w;
+ blendFactor.z = pixel.w;
break;
case Context::BLEND_INVDESTALPHA:
- blendFactor.r = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.a;
- blendFactor.g = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.a;
- blendFactor.b = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.a;
+ blendFactor.x = Float4(1.0f) - pixel.w;
+ blendFactor.y = Float4(1.0f) - pixel.w;
+ blendFactor.z = Float4(1.0f) - pixel.w;
break;
case Context::BLEND_SRCALPHASAT:
- blendFactor.r = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.a;
- blendFactor.r = Min(blendFactor.r, oC.a);
- blendFactor.g = blendFactor.r;
- blendFactor.b = blendFactor.r;
+ blendFactor.x = Float4(1.0f) - pixel.w;
+ blendFactor.x = Min(blendFactor.x, oC.w);
+ blendFactor.y = blendFactor.x;
+ blendFactor.z = blendFactor.x;
break;
case Context::BLEND_CONSTANT:
- blendFactor.r = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[0]));
- blendFactor.g = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[1]));
- blendFactor.b = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[2]));
+ 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]));
break;
case Context::BLEND_INVCONSTANT:
- blendFactor.r = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[0]));
- blendFactor.g = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[1]));
- blendFactor.b = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[2]));
+ 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]));
break;
default:
ASSERT(false);
}
}
- void PixelRoutine::blendFactorAlpha(Registers &r, const Color4f &blendFactor, const Color4f &oC, const Color4f &pixel, Context::BlendFactor blendFactorAlphaActive)
+ void PixelRoutine::blendFactorAlpha(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, Context::BlendFactor blendFactorAlphaActive)
{
switch(blendFactorAlphaActive)
{
@@ -3003,44 +3035,44 @@
// Optimized
break;
case Context::BLEND_SOURCE:
- blendFactor.a = oC.a;
+ blendFactor.w = oC.w;
break;
case Context::BLEND_INVSOURCE:
- blendFactor.a = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.a;
+ blendFactor.w = Float4(1.0f) - oC.w;
break;
case Context::BLEND_DEST:
- blendFactor.a = pixel.a;
+ blendFactor.w = pixel.w;
break;
case Context::BLEND_INVDEST:
- blendFactor.a = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.a;
+ blendFactor.w = Float4(1.0f) - pixel.w;
break;
case Context::BLEND_SOURCEALPHA:
- blendFactor.a = oC.a;
+ blendFactor.w = oC.w;
break;
case Context::BLEND_INVSOURCEALPHA:
- blendFactor.a = Float4(1.0f, 1.0f, 1.0f, 1.0f) - oC.a;
+ blendFactor.w = Float4(1.0f) - oC.w;
break;
case Context::BLEND_DESTALPHA:
- blendFactor.a = pixel.a;
+ blendFactor.w = pixel.w;
break;
case Context::BLEND_INVDESTALPHA:
- blendFactor.a = Float4(1.0f, 1.0f, 1.0f, 1.0f) - pixel.a;
+ blendFactor.w = Float4(1.0f) - pixel.w;
break;
case Context::BLEND_SRCALPHASAT:
- blendFactor.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ blendFactor.w = Float4(1.0f);
break;
case Context::BLEND_CONSTANT:
- blendFactor.a = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[3]));
+ blendFactor.w = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.blendConstant4F[3]));
break;
case Context::BLEND_INVCONSTANT:
- blendFactor.a = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[3]));
+ blendFactor.w = *Pointer<Float4>(r.data + OFFSET(DrawData,factor.invBlendConstant4F[3]));
break;
default:
ASSERT(false);
}
}
- void PixelRoutine::alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Color4f &oC, Int &x)
+ void PixelRoutine::alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4f &oC, Int &x)
{
if(!state.alphaBlendActive)
{
@@ -3048,9 +3080,9 @@
}
Pointer<Byte> buffer;
- Color4f pixel;
+ Vector4f pixel;
- Color4i color;
+ Vector4i color;
Short4 c01;
Short4 c23;
@@ -3062,126 +3094,126 @@
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
- color.b = c01;
- color.g = c01;
- color.b = UnpackLow(As<Byte8>(color.b), As<Byte8>(c23));
- color.g = UnpackHigh(As<Byte8>(color.g), As<Byte8>(c23));
- color.r = color.b;
- color.b = UnpackLow(As<Byte8>(color.b), As<Byte8>(color.g));
- color.r = UnpackHigh(As<Byte8>(color.r), As<Byte8>(color.g));
- color.g = color.b;
- color.a = color.r;
- color.r = UnpackLow(As<Byte8>(color.r), As<Byte8>(color.r));
- color.g = UnpackHigh(As<Byte8>(color.g), As<Byte8>(color.g));
- color.b = UnpackLow(As<Byte8>(color.b), As<Byte8>(color.b));
- color.a = UnpackHigh(As<Byte8>(color.a), As<Byte8>(color.a));
+ color.z = c01;
+ color.y = c01;
+ color.z = UnpackLow(As<Byte8>(color.z), As<Byte8>(c23));
+ color.y = UnpackHigh(As<Byte8>(color.y), As<Byte8>(c23));
+ color.x = color.z;
+ color.z = UnpackLow(As<Byte8>(color.z), As<Byte8>(color.y));
+ color.x = UnpackHigh(As<Byte8>(color.x), As<Byte8>(color.y));
+ color.y = color.z;
+ color.w = color.x;
+ color.x = UnpackLow(As<Byte8>(color.x), As<Byte8>(color.x));
+ color.y = UnpackHigh(As<Byte8>(color.y), As<Byte8>(color.y));
+ color.z = UnpackLow(As<Byte8>(color.z), As<Byte8>(color.z));
+ color.w = UnpackHigh(As<Byte8>(color.w), As<Byte8>(color.w));
- pixel.r = convertUnsigned16(As<UShort4>(color.r));
- pixel.g = convertUnsigned16(As<UShort4>(color.g));
- pixel.b = convertUnsigned16(As<UShort4>(color.b));
- pixel.a = convertUnsigned16(As<UShort4>(color.a));
+ pixel.x = convertUnsigned16(As<UShort4>(color.x));
+ pixel.y = convertUnsigned16(As<UShort4>(color.y));
+ pixel.z = convertUnsigned16(As<UShort4>(color.z));
+ pixel.w = convertUnsigned16(As<UShort4>(color.w));
break;
case FORMAT_X8R8G8B8:
buffer = cBuffer + 4 * x;
c01 = *Pointer<Short4>(buffer);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
c23 = *Pointer<Short4>(buffer);
- color.b = c01;
- color.g = c01;
- color.b = UnpackLow(As<Byte8>(color.b), As<Byte8>(c23));
- color.g = UnpackHigh(As<Byte8>(color.g), As<Byte8>(c23));
- color.r = color.b;
- color.b = UnpackLow(As<Byte8>(color.b), As<Byte8>(color.g));
- color.r = UnpackHigh(As<Byte8>(color.r), As<Byte8>(color.g));
- color.g = color.b;
- color.r = UnpackLow(As<Byte8>(color.r), As<Byte8>(color.r));
- color.g = UnpackHigh(As<Byte8>(color.g), As<Byte8>(color.g));
- color.b = UnpackLow(As<Byte8>(color.b), As<Byte8>(color.b));
+ color.z = c01;
+ color.y = c01;
+ color.z = UnpackLow(As<Byte8>(color.z), As<Byte8>(c23));
+ color.y = UnpackHigh(As<Byte8>(color.y), As<Byte8>(c23));
+ color.x = color.z;
+ color.z = UnpackLow(As<Byte8>(color.z), As<Byte8>(color.y));
+ color.x = UnpackHigh(As<Byte8>(color.x), As<Byte8>(color.y));
+ color.y = color.z;
+ color.x = UnpackLow(As<Byte8>(color.x), As<Byte8>(color.x));
+ color.y = UnpackHigh(As<Byte8>(color.y), As<Byte8>(color.y));
+ color.z = UnpackLow(As<Byte8>(color.z), As<Byte8>(color.z));
- pixel.r = convertUnsigned16(As<UShort4>(color.r));
- pixel.g = convertUnsigned16(As<UShort4>(color.g));
- pixel.b = convertUnsigned16(As<UShort4>(color.b));
- pixel.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ pixel.x = convertUnsigned16(As<UShort4>(color.x));
+ pixel.y = convertUnsigned16(As<UShort4>(color.y));
+ pixel.z = convertUnsigned16(As<UShort4>(color.z));
+ pixel.w = Float4(1.0f);
break;
case FORMAT_A8G8R8B8Q:
UNIMPLEMENTED();
- // UnpackLow(pixel.b, qword_ptr [cBuffer+8*x+0]);
- // UnpackHigh(pixel.r, qword_ptr [cBuffer+8*x+0]);
- // UnpackLow(pixel.g, qword_ptr [cBuffer+8*x+8]);
- // UnpackHigh(pixel.a, qword_ptr [cBuffer+8*x+8]);
+ // UnpackLow(pixel.z, qword_ptr [cBuffer+8*x+0]);
+ // UnpackHigh(pixel.x, qword_ptr [cBuffer+8*x+0]);
+ // UnpackLow(pixel.y, qword_ptr [cBuffer+8*x+8]);
+ // UnpackHigh(pixel.w, qword_ptr [cBuffer+8*x+8]);
break;
case FORMAT_X8G8R8B8Q:
UNIMPLEMENTED();
- // UnpackLow(pixel.b, qword_ptr [cBuffer+8*x+0]);
- // UnpackHigh(pixel.r, qword_ptr [cBuffer+8*x+0]);
- // UnpackLow(pixel.g, qword_ptr [cBuffer+8*x+8]);
- // pixel.a = Short4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF);
+ // UnpackLow(pixel.z, qword_ptr [cBuffer+8*x+0]);
+ // UnpackHigh(pixel.x, qword_ptr [cBuffer+8*x+0]);
+ // UnpackLow(pixel.y, qword_ptr [cBuffer+8*x+8]);
+ // pixel.w = Short4(0xFFFFu);
break;
case FORMAT_A16B16G16R16:
buffer = cBuffer;
- color.r = *Pointer<Short4>(buffer + 8 * x);
- color.g = *Pointer<Short4>(buffer + 8 * x + 8);
+ color.x = *Pointer<Short4>(buffer + 8 * x);
+ color.y = *Pointer<Short4>(buffer + 8 * x + 8);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
- color.b = *Pointer<Short4>(buffer + 8 * x);
- color.a = *Pointer<Short4>(buffer + 8 * x + 8);
+ color.z = *Pointer<Short4>(buffer + 8 * x);
+ color.w = *Pointer<Short4>(buffer + 8 * x + 8);
- transpose4x4(color.r, color.g, color.b, color.a);
+ transpose4x4(color.x, color.y, color.z, color.w);
- pixel.r = convertUnsigned16(As<UShort4>(color.r));
- pixel.g = convertUnsigned16(As<UShort4>(color.g));
- pixel.b = convertUnsigned16(As<UShort4>(color.b));
- pixel.a = convertUnsigned16(As<UShort4>(color.a));
+ pixel.x = convertUnsigned16(As<UShort4>(color.x));
+ pixel.y = convertUnsigned16(As<UShort4>(color.y));
+ pixel.z = convertUnsigned16(As<UShort4>(color.z));
+ pixel.w = convertUnsigned16(As<UShort4>(color.w));
break;
case FORMAT_G16R16:
buffer = cBuffer;
- color.r = *Pointer<Short4>(buffer + 4 * x);
+ color.x = *Pointer<Short4>(buffer + 4 * x);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
- color.g = *Pointer<Short4>(buffer + 4 * x);
- color.b = color.r;
- color.r = As<Short4>(UnpackLow(color.r, color.g));
- color.b = As<Short4>(UnpackHigh(color.b, color.g));
- color.g = color.b;
- color.r = As<Short4>(UnpackLow(color.r, color.b));
- color.g = As<Short4>(UnpackHigh(color.g, color.b));
+ color.y = *Pointer<Short4>(buffer + 4 * x);
+ color.z = color.x;
+ color.x = As<Short4>(UnpackLow(color.x, color.y));
+ color.z = As<Short4>(UnpackHigh(color.z, color.y));
+ color.y = color.z;
+ color.x = As<Short4>(UnpackLow(color.x, color.z));
+ color.y = As<Short4>(UnpackHigh(color.y, color.z));
- pixel.r = convertUnsigned16(As<UShort4>(color.r));
- pixel.g = convertUnsigned16(As<UShort4>(color.g));
- pixel.b = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- pixel.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ pixel.x = convertUnsigned16(As<UShort4>(color.x));
+ pixel.y = convertUnsigned16(As<UShort4>(color.y));
+ pixel.z = Float4(1.0f);
+ pixel.w = Float4(1.0f);
break;
case FORMAT_R32F:
buffer = cBuffer;
// FIXME: movlps
- pixel.r.x = *Pointer<Float>(buffer + 4 * x + 0);
- pixel.r.y = *Pointer<Float>(buffer + 4 * x + 4);
+ 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]));
// FIXME: movhps
- pixel.r.z = *Pointer<Float>(buffer + 4 * x + 0);
- pixel.r.w = *Pointer<Float>(buffer + 4 * x + 4);
- pixel.g = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- pixel.b = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- pixel.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ pixel.x.z = *Pointer<Float>(buffer + 4 * x + 0);
+ pixel.x.w = *Pointer<Float>(buffer + 4 * x + 4);
+ pixel.y = Float4(1.0f);
+ pixel.z = Float4(1.0f);
+ pixel.w = Float4(1.0f);
break;
case FORMAT_G32R32F:
buffer = cBuffer;
- pixel.r = *Pointer<Float4>(buffer + 8 * x, 16);
+ pixel.x = *Pointer<Float4>(buffer + 8 * x, 16);
buffer += *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
- pixel.g = *Pointer<Float4>(buffer + 8 * x, 16);
- pixel.b = pixel.r;
- pixel.r = ShuffleLowHigh(pixel.r, pixel.g, 0x88);
- pixel.b = ShuffleLowHigh(pixel.b, pixel.g, 0xDD);
- pixel.g = pixel.b;
- pixel.b = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- pixel.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ pixel.y = *Pointer<Float4>(buffer + 8 * x, 16);
+ pixel.z = pixel.x;
+ pixel.x = ShuffleLowHigh(pixel.x, pixel.y, 0x88);
+ pixel.z = ShuffleLowHigh(pixel.z, pixel.y, 0xDD);
+ pixel.y = pixel.z;
+ pixel.z = Float4(1.0f);
+ pixel.w = Float4(1.0f);
break;
case FORMAT_A32B32G32R32F:
buffer = cBuffer;
- pixel.r = *Pointer<Float4>(buffer + 16 * x, 16);
- pixel.g = *Pointer<Float4>(buffer + 16 * x + 16, 16);
+ 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]));
- pixel.b = *Pointer<Float4>(buffer + 16 * x, 16);
- pixel.a = *Pointer<Float4>(buffer + 16 * x + 16, 16);
- transpose4x4(pixel.r, pixel.g, pixel.b, pixel.a);
+ 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);
break;
default:
ASSERT(false);
@@ -3189,71 +3221,71 @@
if(postBlendSRGB && state.writeSRGB)
{
- sRGBtoLinear(pixel.r);
- sRGBtoLinear(pixel.g);
- sRGBtoLinear(pixel.b);
+ sRGBtoLinear(pixel.x);
+ sRGBtoLinear(pixel.y);
+ sRGBtoLinear(pixel.z);
}
// Final Color = ObjectColor * SourceBlendFactor + PixelColor * DestinationBlendFactor
- Color4f sourceFactor;
- Color4f destFactor;
+ Vector4f sourceFactor;
+ Vector4f destFactor;
blendFactor(r, sourceFactor, oC, pixel, (Context::BlendFactor)state.sourceBlendFactor);
blendFactor(r, destFactor, oC, pixel, (Context::BlendFactor)state.destBlendFactor);
if(state.sourceBlendFactor != Context::BLEND_ONE && state.sourceBlendFactor != Context::BLEND_ZERO)
{
- oC.r *= sourceFactor.r;
- oC.g *= sourceFactor.g;
- oC.b *= sourceFactor.b;
+ oC.x *= sourceFactor.x;
+ oC.y *= sourceFactor.y;
+ oC.z *= sourceFactor.z;
}
if(state.destBlendFactor != Context::BLEND_ONE && state.destBlendFactor != Context::BLEND_ZERO)
{
- pixel.r *= destFactor.r;
- pixel.g *= destFactor.g;
- pixel.b *= destFactor.b;
+ pixel.x *= destFactor.x;
+ pixel.y *= destFactor.y;
+ pixel.z *= destFactor.z;
}
switch(state.blendOperation)
{
case Context::BLENDOP_ADD:
- oC.r += pixel.r;
- oC.g += pixel.g;
- oC.b += pixel.b;
+ oC.x += pixel.x;
+ oC.y += pixel.y;
+ oC.z += pixel.z;
break;
case Context::BLENDOP_SUB:
- oC.r -= pixel.r;
- oC.g -= pixel.g;
- oC.b -= pixel.b;
+ oC.x -= pixel.x;
+ oC.y -= pixel.y;
+ oC.z -= pixel.z;
break;
case Context::BLENDOP_INVSUB:
- oC.r = pixel.r - oC.r;
- oC.g = pixel.g - oC.g;
- oC.b = pixel.b - oC.b;
+ oC.x = pixel.x - oC.x;
+ oC.y = pixel.y - oC.y;
+ oC.z = pixel.z - oC.z;
break;
case Context::BLENDOP_MIN:
- oC.r = Min(oC.r, pixel.r);
- oC.g = Min(oC.g, pixel.g);
- oC.b = Min(oC.b, pixel.b);
+ oC.x = Min(oC.x, pixel.x);
+ oC.y = Min(oC.y, pixel.y);
+ oC.z = Min(oC.z, pixel.z);
break;
case Context::BLENDOP_MAX:
- oC.r = Max(oC.r, pixel.r);
- oC.g = Max(oC.g, pixel.g);
- oC.b = Max(oC.b, pixel.b);
+ oC.x = Max(oC.x, pixel.x);
+ oC.y = Max(oC.y, pixel.y);
+ oC.z = Max(oC.z, pixel.z);
break;
case Context::BLENDOP_SOURCE:
// No operation
break;
case Context::BLENDOP_DEST:
- oC.r = pixel.r;
- oC.g = pixel.g;
- oC.b = pixel.b;
+ oC.x = pixel.x;
+ oC.y = pixel.y;
+ oC.z = pixel.z;
break;
case Context::BLENDOP_NULL:
- oC.r = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- oC.g = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- oC.b = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ oC.x = Float4(0.0f);
+ oC.y = Float4(0.0f);
+ oC.z = Float4(0.0f);
break;
default:
ASSERT(false);
@@ -3264,54 +3296,54 @@
if(state.sourceBlendFactorAlpha != Context::BLEND_ONE && state.sourceBlendFactorAlpha != Context::BLEND_ZERO)
{
- oC.a *= sourceFactor.a;
+ oC.w *= sourceFactor.w;
}
if(state.destBlendFactorAlpha != Context::BLEND_ONE && state.destBlendFactorAlpha != Context::BLEND_ZERO)
{
- pixel.a *= destFactor.a;
+ pixel.w *= destFactor.w;
}
switch(state.blendOperationAlpha)
{
case Context::BLENDOP_ADD:
- oC.a += pixel.a;
+ oC.w += pixel.w;
break;
case Context::BLENDOP_SUB:
- oC.a -= pixel.a;
+ oC.w -= pixel.w;
break;
case Context::BLENDOP_INVSUB:
- pixel.a -= oC.a;
- oC.a = pixel.a;
+ pixel.w -= oC.w;
+ oC.w = pixel.w;
break;
case Context::BLENDOP_MIN:
- oC.a = Min(oC.a, pixel.a);
+ oC.w = Min(oC.w, pixel.w);
break;
case Context::BLENDOP_MAX:
- oC.a = Max(oC.a, pixel.a);
+ oC.w = Max(oC.w, pixel.w);
break;
case Context::BLENDOP_SOURCE:
// No operation
break;
case Context::BLENDOP_DEST:
- oC.a = pixel.a;
+ oC.w = pixel.w;
break;
case Context::BLENDOP_NULL:
- oC.a = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ oC.w = Float4(0.0f);
break;
default:
ASSERT(false);
}
}
- void PixelRoutine::writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Color4f &oC, Int &sMask, Int &zMask, Int &cMask)
+ void PixelRoutine::writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &x, Vector4f &oC, Int &sMask, Int &zMask, Int &cMask)
{
if(!state.colorWriteActive(index))
{
return;
}
- Color4i color;
+ Vector4i color;
switch(state.targetFormat[index])
{
@@ -3325,13 +3357,13 @@
case FORMAT_R32F:
break;
case FORMAT_G32R32F:
- oC.b = oC.r;
- oC.r = UnpackLow(oC.r, oC.g);
- oC.b = UnpackHigh(oC.b, oC.g);
- oC.g = oC.b;
+ oC.z = oC.x;
+ oC.x = UnpackLow(oC.x, oC.y);
+ oC.z = UnpackHigh(oC.z, oC.y);
+ oC.y = oC.z;
break;
case FORMAT_A32B32G32R32F:
- transpose4x4(oC.r, oC.g, oC.b, oC.a);
+ transpose4x4(oC.x, oC.y, oC.z, oC.w);
break;
default:
ASSERT(false);
@@ -3375,19 +3407,19 @@
value.z = *Pointer<Float>(buffer + 0);
value.w = *Pointer<Float>(buffer + 4);
- oC.r = As<Float4>(As<Int4>(oC.r) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X) + xMask * 16, 16));
+ 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.r = As<Float4>(As<Int4>(oC.r) | As<Int4>(value));
+ oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(value));
// FIXME: movhps
- *Pointer<Float>(buffer + 0) = oC.r.z;
- *Pointer<Float>(buffer + 4) = oC.r.w;
+ *Pointer<Float>(buffer + 0) = oC.x.z;
+ *Pointer<Float>(buffer + 4) = oC.x.w;
buffer -= *Pointer<Int>(r.data + OFFSET(DrawData,colorPitchB[index]));
// FIXME: movlps
- *Pointer<Float>(buffer + 0) = oC.r.x;
- *Pointer<Float>(buffer + 4) = oC.r.y;
+ *Pointer<Float>(buffer + 0) = oC.x.x;
+ *Pointer<Float>(buffer + 4) = oC.x.y;
}
break;
case FORMAT_G32R32F:
@@ -3398,15 +3430,15 @@
if((rgbaWriteMask & 0x00000003) != 0x00000003)
{
Float4 masked = value;
- oC.r = As<Float4>(As<Int4>(oC.r) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD01X[rgbaWriteMask & 0x3][0])));
+ 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.r = As<Float4>(As<Int4>(oC.r) | As<Int4>(masked));
+ oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(masked));
}
- oC.r = As<Float4>(As<Int4>(oC.r) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskQ01X) + xMask * 16, 16));
+ 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.r = As<Float4>(As<Int4>(oC.r) | As<Int4>(value));
- *Pointer<Float4>(buffer) = oC.r;
+ 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]));
@@ -3417,15 +3449,15 @@
Float4 masked;
masked = value;
- oC.g = As<Float4>(As<Int4>(oC.g) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD01X[rgbaWriteMask & 0x3][0])));
+ 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.g = As<Float4>(As<Int4>(oC.g) | As<Int4>(masked));
+ oC.y = As<Float4>(As<Int4>(oC.y) | As<Int4>(masked));
}
- oC.g = As<Float4>(As<Int4>(oC.g) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskQ23X) + xMask * 16, 16));
+ 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.g = As<Float4>(As<Int4>(oC.g) | As<Int4>(value));
- *Pointer<Float4>(buffer) = oC.g;
+ oC.y = As<Float4>(As<Int4>(oC.y) | As<Int4>(value));
+ *Pointer<Float4>(buffer) = oC.y;
break;
case FORMAT_A32B32G32R32F:
buffer = cBuffer + 16 * x;
@@ -3436,15 +3468,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
- oC.r = As<Float4>(As<Int4>(oC.r) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
+ 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.r = As<Float4>(As<Int4>(oC.r) | As<Int4>(masked));
+ oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(masked));
}
- oC.r = As<Float4>(As<Int4>(oC.r) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX0X) + xMask * 16, 16));
+ 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.r = As<Float4>(As<Int4>(oC.r) | As<Int4>(value));
- *Pointer<Float4>(buffer, 16) = oC.r;
+ oC.x = As<Float4>(As<Int4>(oC.x) | As<Int4>(value));
+ *Pointer<Float4>(buffer, 16) = oC.x;
}
{
@@ -3453,15 +3485,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
- oC.g = As<Float4>(As<Int4>(oC.g) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
+ 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.g = As<Float4>(As<Int4>(oC.g) | As<Int4>(masked));
+ oC.y = As<Float4>(As<Int4>(oC.y) | As<Int4>(masked));
}
- oC.g = As<Float4>(As<Int4>(oC.g) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX1X) + xMask * 16, 16));
+ 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.g = As<Float4>(As<Int4>(oC.g) | As<Int4>(value));
- *Pointer<Float4>(buffer + 16, 16) = oC.g;
+ 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]));
@@ -3472,15 +3504,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
- oC.b = As<Float4>(As<Int4>(oC.b) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
+ 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.b = As<Float4>(As<Int4>(oC.b) | As<Int4>(masked));
+ oC.z = As<Float4>(As<Int4>(oC.z) | As<Int4>(masked));
}
- oC.b = As<Float4>(As<Int4>(oC.b) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX2X) + xMask * 16, 16));
+ 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.b = As<Float4>(As<Int4>(oC.b) | As<Int4>(value));
- *Pointer<Float4>(buffer, 16) = oC.b;
+ oC.z = As<Float4>(As<Int4>(oC.z) | As<Int4>(value));
+ *Pointer<Float4>(buffer, 16) = oC.z;
}
{
@@ -3489,15 +3521,15 @@
if(rgbaWriteMask != 0x0000000F)
{
Float4 masked = value;
- oC.a = As<Float4>(As<Int4>(oC.a) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskD4X[rgbaWriteMask][0])));
+ 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.a = As<Float4>(As<Int4>(oC.a) | As<Int4>(masked));
+ oC.w = As<Float4>(As<Int4>(oC.w) | As<Int4>(masked));
}
- oC.a = As<Float4>(As<Int4>(oC.a) & *Pointer<Int4>(r.constants + OFFSET(Constants,maskX3X) + xMask * 16, 16));
+ 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.a = As<Float4>(As<Int4>(oC.a) | As<Int4>(value));
- *Pointer<Float4>(buffer + 16, 16) = oC.a;
+ oC.w = As<Float4>(As<Int4>(oC.w) | As<Int4>(value));
+ *Pointer<Float4>(buffer + 16, 16) = oC.w;
}
break;
default:
@@ -3508,88 +3540,91 @@
void PixelRoutine::ps_1_x(Registers &r, Int cMask[4])
{
int pad = 0; // Count number of texm3x3pad instructions
- Color4i dPairing; // Destination for first pairing instruction
+ Vector4i dPairing; // Destination for first pairing instruction
- for(int i = 0; i < pixelShader->getLength(); i++)
+ for(int i = 0; i < shader->getLength(); i++)
{
- const ShaderInstruction *instruction = pixelShader->getInstruction(i);
- Op::Opcode opcode = instruction->getOpcode();
+ const Shader::Instruction *instruction = shader->getInstruction(i);
+ Shader::Opcode opcode = instruction->opcode;
// #ifndef NDEBUG // FIXME: Centralize debug output control
- // pixelShader->printInstruction(i, "debug.txt");
+ // shader->printInstruction(i, "debug.txt");
// #endif
- if(opcode == Op::OPCODE_DCL || opcode == Op::OPCODE_DEF || opcode == Op::OPCODE_DEFI || opcode == Op::OPCODE_DEFB)
+ if(opcode == Shader::OPCODE_DCL || opcode == Shader::OPCODE_DEF || opcode == Shader::OPCODE_DEFI || opcode == Shader::OPCODE_DEFB)
{
continue;
}
- const Dst &dst = instruction->getDestinationParameter();
- const Src &src0 = instruction->getSourceParameter(0);
- const Src &src1 = instruction->getSourceParameter(1);
- const Src &src2 = instruction->getSourceParameter(2);
- const Src &src3 = instruction->getSourceParameter(3);
+ const Dst &dst = instruction->dst;
+ const Src &src0 = instruction->src[0];
+ const Src &src1 = instruction->src[1];
+ const Src &src2 = instruction->src[2];
- bool pairing = i + 1 < pixelShader->getLength() && pixelShader->getInstruction(i + 1)->isCoissue(); // First instruction of pair
- bool coissue = instruction->isCoissue(); // Second instruction of pair
+ unsigned short version = shader->getVersion();
+ bool pairing = i + 1 < shader->getLength() && shader->getInstruction(i + 1)->coissue; // First instruction of pair
+ bool coissue = instruction->coissue; // Second instruction of pair
- Color4i d;
- Color4i s0;
- Color4i s1;
- Color4i s2;
- Color4i s3;
+ Vector4i d;
+ Vector4i s0;
+ Vector4i s1;
+ Vector4i s2;
- if(src0.type != Src::PARAMETER_VOID) s0 = regi(r, src0);
- if(src1.type != Src::PARAMETER_VOID) s1 = regi(r, src1);
- if(src2.type != Src::PARAMETER_VOID) s2 = regi(r, src2);
- if(src3.type != Src::PARAMETER_VOID) s3 = regi(r, src3);
+ if(src0.type != Shader::PARAMETER_VOID) s0 = regi(r, src0);
+ if(src1.type != Shader::PARAMETER_VOID) s1 = regi(r, src1);
+ if(src2.type != Shader::PARAMETER_VOID) s2 = regi(r, src2);
+
+ Float4 u = version < 0x0104 ? r.vf[2 + dst.index].x : r.vf[2 + src0.index].x;
+ Float4 v = version < 0x0104 ? r.vf[2 + dst.index].y : r.vf[2 + src0.index].y;
+ Float4 s = version < 0x0104 ? r.vf[2 + dst.index].z : r.vf[2 + src0.index].z;
+ Float4 t = version < 0x0104 ? r.vf[2 + dst.index].w : r.vf[2 + src0.index].w;
switch(opcode)
{
- case Op::OPCODE_PS_1_0: break;
- case Op::OPCODE_PS_1_1: break;
- case Op::OPCODE_PS_1_2: break;
- case Op::OPCODE_PS_1_3: break;
- case Op::OPCODE_PS_1_4: break;
+ case Shader::OPCODE_PS_1_0: break;
+ case Shader::OPCODE_PS_1_1: break;
+ case Shader::OPCODE_PS_1_2: break;
+ case Shader::OPCODE_PS_1_3: break;
+ case Shader::OPCODE_PS_1_4: break;
- case Op::OPCODE_DEF: break;
+ case Shader::OPCODE_DEF: break;
- case Op::OPCODE_NOP: break;
- case Op::OPCODE_MOV: MOV(d, s0); break;
- case Op::OPCODE_ADD: ADD(d, s0, s1); break;
- case Op::OPCODE_SUB: SUB(d, s0, s1); break;
- case Op::OPCODE_MAD: MAD(d, s0, s1, s2); break;
- case Op::OPCODE_MUL: MUL(d, s0, s1); break;
- case Op::OPCODE_DP3: DP3(d, s0, s1); break;
- case Op::OPCODE_DP4: DP4(d, s0, s1); break;
- case Op::OPCODE_LRP: LRP(d, s0, s1, s2); break;
- case Op::OPCODE_TEXCOORD:
- if(pixelShader->getVersion() < 0x0104)
+ case Shader::OPCODE_NOP: break;
+ case Shader::OPCODE_MOV: MOV(d, s0); break;
+ case Shader::OPCODE_ADD: ADD(d, s0, s1); break;
+ case Shader::OPCODE_SUB: SUB(d, s0, s1); break;
+ case Shader::OPCODE_MAD: MAD(d, s0, s1, s2); break;
+ case Shader::OPCODE_MUL: MUL(d, s0, s1); break;
+ case Shader::OPCODE_DP3: DP3(d, s0, s1); break;
+ case Shader::OPCODE_DP4: DP4(d, s0, s1); break;
+ case Shader::OPCODE_LRP: LRP(d, s0, s1, s2); break;
+ case Shader::OPCODE_TEXCOORD:
+ if(version < 0x0104)
{
- TEXCOORD(d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index);
+ TEXCOORD(d, u, v, s, dst.index);
}
else
{
if((src0.swizzle & 0x30) == 0x20) // .xyz
{
- TEXCRD(d, Float4(r.vx[2 + src0.index]), Float4(r.vy[2 + src0.index]), Float4(r.vz[2 + src0.index]), src0.index, src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DZ || src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DW);
+ TEXCRD(d, u, v, s, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
else // .xyw
{
- TEXCRD(d, Float4(r.vx[2 + src0.index]), Float4(r.vy[2 + src0.index]), Float4(r.vw[2 + src0.index]), src0.index, src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DZ || src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DW);
+ TEXCRD(d, u, v, t, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
}
break;
- case Op::OPCODE_TEXKILL:
- if(pixelShader->getVersion() < 0x0104)
+ case Shader::OPCODE_TEXKILL:
+ if(version < 0x0104)
{
- TEXKILL(cMask, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]));
+ TEXKILL(cMask, u, v, s);
}
- else if(pixelShader->getVersion() == 0x0104)
+ else if(version == 0x0104)
{
- if(dst.type == Dst::PARAMETER_TEXTURE)
+ if(dst.type == Shader::PARAMETER_TEXTURE)
{
- TEXKILL(cMask, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]));
+ TEXKILL(cMask, u, v, s);
}
else
{
@@ -3598,92 +3633,92 @@
}
else ASSERT(false);
break;
- case Op::OPCODE_TEX:
- if(pixelShader->getVersion() < 0x0104)
+ case Shader::OPCODE_TEX:
+ if(version < 0x0104)
{
- TEX(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index, false);
+ TEX(r, d, u, v, s, dst.index, false);
}
- else if(pixelShader->getVersion() == 0x0104)
+ else if(version == 0x0104)
{
- if(src0.type == Src::PARAMETER_TEXTURE)
+ if(src0.type == Shader::PARAMETER_TEXTURE)
{
if((src0.swizzle & 0x30) == 0x20) // .xyz
{
- TEX(r, d, Float4(r.vx[2 + src0.index]), Float4(r.vy[2 + src0.index]), Float4(r.vz[2 + src0.index]), dst.index, src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DZ || src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DW);
+ TEX(r, d, u, v, s, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
else // .xyw
{
- TEX(r, d, Float4(r.vx[2 + src0.index]), Float4(r.vy[2 + src0.index]), Float4(r.vw[2 + src0.index]), dst.index, src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DZ || src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DW);
+ TEX(r, d, u, v, t, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
}
else
{
- TEXLD(r, d, s0, dst.index, src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DZ || src0.modifier == ShaderInstruction::SourceParameter::MODIFIER_DW);
+ TEXLD(r, d, s0, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW);
}
}
else ASSERT(false);
break;
- case Op::OPCODE_TEXBEM: TEXBEM(r, d, s0, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index); break;
- case Op::OPCODE_TEXBEML: TEXBEML(r, d, s0, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index); break;
- case Op::OPCODE_TEXREG2AR: TEXREG2AR(r, d, s0, dst.index); break;
- case Op::OPCODE_TEXREG2GB: TEXREG2GB(r, d, s0, dst.index); break;
- case Op::OPCODE_TEXM3X2PAD: TEXM3X2PAD(r, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), s0, 0, src0.modifier == Src::MODIFIER_SIGN); break;
- case Op::OPCODE_TEXM3X2TEX: TEXM3X2TEX(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index, s0, src0.modifier == Src::MODIFIER_SIGN); break;
- case Op::OPCODE_TEXM3X3PAD: TEXM3X3PAD(r, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), s0, pad++ % 2, src0.modifier == Src::MODIFIER_SIGN); break;
- case Op::OPCODE_TEXM3X3TEX: TEXM3X3TEX(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index, s0, src0.modifier == Src::MODIFIER_SIGN); break;
- case Op::OPCODE_TEXM3X3SPEC: TEXM3X3SPEC(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index, s0, s1); break;
- case Op::OPCODE_TEXM3X3VSPEC: TEXM3X3VSPEC(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index, s0); break;
- case Op::OPCODE_CND: CND(d, s0, s1, s2); break;
- case Op::OPCODE_TEXREG2RGB: TEXREG2RGB(r, d, s0, dst.index); break;
- case Op::OPCODE_TEXDP3TEX: TEXDP3TEX(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), dst.index, s0); break;
- case Op::OPCODE_TEXM3X2DEPTH: TEXM3X2DEPTH(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), s0, src0.modifier == Src::MODIFIER_SIGN); break;
- case Op::OPCODE_TEXDP3: TEXDP3(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), s0); break;
- case Op::OPCODE_TEXM3X3: TEXM3X3(r, d, Float4(r.vx[2 + dst.index]), Float4(r.vy[2 + dst.index]), Float4(r.vz[2 + dst.index]), s0, src0.modifier == Src::MODIFIER_SIGN); break;
- case Op::OPCODE_TEXDEPTH: TEXDEPTH(r); break;
- case Op::OPCODE_CMP: CMP(d, s0, s1, s2); break;
- case Op::OPCODE_BEM: BEM(r, d, s0, s1, dst.index); break;
- case Op::OPCODE_PHASE: break;
- case Op::OPCODE_END: 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;
default:
ASSERT(false);
}
- if(dst.type != Dst::PARAMETER_VOID && opcode != Op::OPCODE_TEXKILL)
+ if(dst.type != Shader::PARAMETER_VOID && opcode != Shader::OPCODE_TEXKILL)
{
if(dst.shift > 0)
{
- if(dst.mask & 0x1) {d.r = AddSat(d.r, d.r); if(dst.shift > 1) d.r = AddSat(d.r, d.r); if(dst.shift > 2) d.r = AddSat(d.r, d.r);}
- if(dst.mask & 0x2) {d.g = AddSat(d.g, d.g); if(dst.shift > 1) d.g = AddSat(d.g, d.g); if(dst.shift > 2) d.g = AddSat(d.g, d.g);}
- if(dst.mask & 0x4) {d.b = AddSat(d.b, d.b); if(dst.shift > 1) d.b = AddSat(d.b, d.b); if(dst.shift > 2) d.b = AddSat(d.b, d.b);}
- if(dst.mask & 0x8) {d.a = AddSat(d.a, d.a); if(dst.shift > 1) d.a = AddSat(d.a, d.a); if(dst.shift > 2) d.a = AddSat(d.a, d.a);}
+ if(dst.mask & 0x1) {d.x = AddSat(d.x, d.x); if(dst.shift > 1) d.x = AddSat(d.x, d.x); if(dst.shift > 2) d.x = AddSat(d.x, d.x);}
+ if(dst.mask & 0x2) {d.y = AddSat(d.y, d.y); if(dst.shift > 1) d.y = AddSat(d.y, d.y); if(dst.shift > 2) d.y = AddSat(d.y, d.y);}
+ if(dst.mask & 0x4) {d.z = AddSat(d.z, d.z); if(dst.shift > 1) d.z = AddSat(d.z, d.z); if(dst.shift > 2) d.z = AddSat(d.z, d.z);}
+ if(dst.mask & 0x8) {d.w = AddSat(d.w, d.w); if(dst.shift > 1) d.w = AddSat(d.w, d.w); if(dst.shift > 2) d.w = AddSat(d.w, d.w);}
}
else if(dst.shift < 0)
{
- if(dst.mask & 0x1) d.r = d.r >> -dst.shift;
- if(dst.mask & 0x2) d.g = d.g >> -dst.shift;
- if(dst.mask & 0x4) d.b = d.b >> -dst.shift;
- if(dst.mask & 0x8) d.a = d.a >> -dst.shift;
+ if(dst.mask & 0x1) d.x = d.x >> -dst.shift;
+ if(dst.mask & 0x2) d.y = d.y >> -dst.shift;
+ if(dst.mask & 0x4) d.z = d.z >> -dst.shift;
+ if(dst.mask & 0x8) d.w = d.w >> -dst.shift;
}
if(dst.saturate)
{
- if(dst.mask & 0x1) {d.r = Min(d.r, Short4(0x1000, 0x1000, 0x1000, 0x1000)); d.r = Max(d.r, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
- if(dst.mask & 0x2) {d.g = Min(d.g, Short4(0x1000, 0x1000, 0x1000, 0x1000)); d.g = Max(d.g, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
- if(dst.mask & 0x4) {d.b = Min(d.b, Short4(0x1000, 0x1000, 0x1000, 0x1000)); d.b = Max(d.b, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
- if(dst.mask & 0x8) {d.a = Min(d.a, Short4(0x1000, 0x1000, 0x1000, 0x1000)); d.a = Max(d.a, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
+ if(dst.mask & 0x1) {d.x = Min(d.x, Short4(0x1000)); d.x = Max(d.x, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
+ if(dst.mask & 0x2) {d.y = Min(d.y, Short4(0x1000)); d.y = Max(d.y, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
+ if(dst.mask & 0x4) {d.z = Min(d.z, Short4(0x1000)); d.z = Max(d.z, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
+ if(dst.mask & 0x8) {d.w = Min(d.w, Short4(0x1000)); d.w = Max(d.w, Short4(0x0000, 0x0000, 0x0000, 0x0000));}
}
if(pairing)
{
- if(dst.mask & 0x1) dPairing.r = d.r;
- if(dst.mask & 0x2) dPairing.g = d.g;
- if(dst.mask & 0x4) dPairing.b = d.b;
- if(dst.mask & 0x8) dPairing.a = d.a;
+ if(dst.mask & 0x1) dPairing.x = d.x;
+ if(dst.mask & 0x2) dPairing.y = d.y;
+ if(dst.mask & 0x4) dPairing.z = d.z;
+ if(dst.mask & 0x8) dPairing.w = d.w;
}
if(coissue)
{
- const Dst &dst = pixelShader->getInstruction(i - 1)->getDestinationParameter();
+ const Dst &dst = shader->getInstruction(i - 1)->dst;
writeDestination(r, dPairing, dst);
}
@@ -3700,47 +3735,60 @@
{
r.enableIndex = 0;
r.stackIndex = 0;
-
- for(int i = 0; i < pixelShader->getLength(); i++)
+
+ bool out[4][4] = {false};
+
+ // Create all call site return blocks up front
+ for(int i = 0; i < shader->getLength(); i++)
{
- const ShaderInstruction *instruction = pixelShader->getInstruction(i);
- Op::Opcode opcode = instruction->getOpcode();
+ const Shader::Instruction *instruction = shader->getInstruction(i);
+ Shader::Opcode opcode = instruction->opcode;
- // #ifndef NDEBUG // FIXME: Centralize debug output control
- // pixelShader->printInstruction(i, "debug.txt");
- // #endif
+ if(opcode == Shader::OPCODE_CALL || opcode == Shader::OPCODE_CALLNZ)
+ {
+ const Dst &dst = instruction->dst;
- if(opcode == Op::OPCODE_DCL || opcode == Op::OPCODE_DEF || opcode == Op::OPCODE_DEFI || opcode == Op::OPCODE_DEFB)
+ ASSERT(callRetBlock[dst.label].size() == dst.callSite);
+ callRetBlock[dst.label].push_back(Nucleus::createBasicBlock());
+ }
+ }
+
+ for(int i = 0; i < shader->getLength(); i++)
+ {
+ const Shader::Instruction *instruction = shader->getInstruction(i);
+ Shader::Opcode opcode = instruction->opcode;
+
+ if(opcode == Shader::OPCODE_DCL || opcode == Shader::OPCODE_DEF || opcode == Shader::OPCODE_DEFI || opcode == Shader::OPCODE_DEFB)
{
continue;
}
- const Dst &dst = instruction->getDestinationParameter();
- const Src &src0 = instruction->getSourceParameter(0);
- const Src &src1 = instruction->getSourceParameter(1);
- const Src &src2 = instruction->getSourceParameter(2);
- const Src &src3 = instruction->getSourceParameter(3);
+ const Dst &dst = instruction->dst;
+ const Src &src0 = instruction->src[0];
+ const Src &src1 = instruction->src[1];
+ const Src &src2 = instruction->src[2];
+ const Src &src3 = instruction->src[3];
- bool predicate = instruction->isPredicate();
- Control control = instruction->getControl();
+ bool predicate = instruction->predicate;
+ Control control = instruction->control;
bool pp = dst.partialPrecision;
- bool project = instruction->isProject();
- bool bias = instruction->isBias();
+ bool project = instruction->project;
+ bool bias = instruction->bias;
- Color4f d;
- Color4f s0;
- Color4f s1;
- Color4f s2;
- Color4f s3;
+ Vector4f d;
+ Vector4f s0;
+ Vector4f s1;
+ Vector4f s2;
+ Vector4f s3;
- if(opcode == Op::OPCODE_TEXKILL)
+ if(opcode == Shader::OPCODE_TEXKILL) // Takes destination as input
{
- if(dst.type == Dst::PARAMETER_TEXTURE)
+ if(dst.type == Shader::PARAMETER_TEXTURE)
{
- d.x = r.vx[2 + dst.index];
- d.y = r.vy[2 + dst.index];
- d.z = r.vz[2 + dst.index];
- d.w = r.vw[2 + dst.index];
+ d.x = r.vf[2 + dst.index].x;
+ d.y = r.vf[2 + dst.index].y;
+ d.z = r.vf[2 + dst.index].z;
+ d.w = r.vf[2 + dst.index].w;
}
else
{
@@ -3748,120 +3796,212 @@
}
}
- if(src0.type != Src::PARAMETER_VOID) s0 = reg(r, src0);
- if(src1.type != Src::PARAMETER_VOID) s1 = reg(r, src1);
- if(src2.type != Src::PARAMETER_VOID) s2 = reg(r, src2);
- if(src3.type != Src::PARAMETER_VOID) s3 = reg(r, src3);
+ if(src0.type != Shader::PARAMETER_VOID) s0 = reg(r, src0);
+ if(src1.type != Shader::PARAMETER_VOID) s1 = reg(r, src1);
+ if(src2.type != Shader::PARAMETER_VOID) s2 = reg(r, src2);
+ if(src3.type != Shader::PARAMETER_VOID) s3 = reg(r, src3);
switch(opcode)
{
- case Op::OPCODE_PS_2_0: break;
- case Op::OPCODE_PS_2_x: break;
- case Op::OPCODE_PS_3_0: break;
- case Op::OPCODE_DEF: break;
- case Op::OPCODE_DCL: break;
- case Op::OPCODE_NOP: break;
- case Op::OPCODE_MOV: mov(d, s0); break;
- case Op::OPCODE_ADD: add(d, s0, s1); break;
- case Op::OPCODE_SUB: sub(d, s0, s1); break;
- case Op::OPCODE_MUL: mul(d, s0, s1); break;
- case Op::OPCODE_MAD: mad(d, s0, s1, s2); break;
- case Op::OPCODE_DP2ADD: dp2add(d, s0, s1, s2); break;
- case Op::OPCODE_DP3: dp3(d, s0, s1); break;
- case Op::OPCODE_DP4: dp4(d, s0, s1); break;
- case Op::OPCODE_CMP: cmp(d, s0, s1, s2); break;
- case Op::OPCODE_FRC: frc(d, s0); break;
- case Op::OPCODE_EXP: exp(d, s0, pp); break;
- case Op::OPCODE_LOG: log(d, s0, pp); break;
- case Op::OPCODE_RCP: rcp(d, s0, pp); break;
- case Op::OPCODE_RSQ: rsq(d, s0, pp); break;
- case Op::OPCODE_MIN: min(d, s0, s1); break;
- case Op::OPCODE_MAX: max(d, s0, s1); break;
- case Op::OPCODE_LRP: lrp(d, s0, s1, s2); break;
- case Op::OPCODE_POW: pow(d, s0, s1, pp); break;
- case Op::OPCODE_CRS: crs(d, s0, s1); break;
- case Op::OPCODE_NRM: nrm(d, s0, pp); break;
- case Op::OPCODE_ABS: abs(d, s0); break;
- case Op::OPCODE_SINCOS: sincos(d, s0, pp); break;
- case Op::OPCODE_M4X4: M4X4(r, d, s0, src1); break;
- case Op::OPCODE_M4X3: M4X3(r, d, s0, src1); break;
- case Op::OPCODE_M3X4: M3X4(r, d, s0, src1); break;
- case Op::OPCODE_M3X3: M3X3(r, d, s0, src1); break;
- case Op::OPCODE_M3X2: M3X2(r, d, s0, src1); break;
- case Op::OPCODE_TEX: TEXLD(r, d, s0, src1, project, bias); break;
- case Op::OPCODE_TEXLDD: TEXLDD(r, d, s0, src1, s2, s3, project, bias); break;
- case Op::OPCODE_TEXLDL: TEXLDL(r, d, s0, src1, project, bias); break;
- case Op::OPCODE_TEXKILL: TEXKILL(cMask, d, dst.mask); break;
- case Op::OPCODE_DSX: DSX(d, s0); break;
- case Op::OPCODE_DSY: DSY(d, s0); break;
- case Op::OPCODE_BREAK: BREAK(r); break;
- case Op::OPCODE_BREAKC: BREAKC(r, s0, s1, control); break;
- case Op::OPCODE_BREAKP: BREAKP(r, src0); break;
- case Op::OPCODE_CALL: CALL(r, dst.index); break;
- case Op::OPCODE_CALLNZ: CALLNZ(r, dst.index, src0); break;
- case Op::OPCODE_ELSE: ELSE(r); break;
- case Op::OPCODE_ENDIF: ENDIF(r); break;
- case Op::OPCODE_ENDLOOP: ENDLOOP(r); break;
- case Op::OPCODE_ENDREP: ENDREP(r); break;
- case Op::OPCODE_IF: IF(r, src0); break;
- case Op::OPCODE_IFC: IFC(r, s0, s1, control); break;
- case Op::OPCODE_LABEL: LABEL(dst.index); break;
- case Op::OPCODE_LOOP: LOOP(r, src1); break;
- case Op::OPCODE_REP: REP(r, src0); break;
- case Op::OPCODE_RET: RET(r); break;
- case Op::OPCODE_SETP: setp(d, s0, s1, control); break;
- case Op::OPCODE_END: break;
+ case Shader::OPCODE_PS_2_0: break;
+ case Shader::OPCODE_PS_2_x: break;
+ case Shader::OPCODE_PS_3_0: break;
+ case Shader::OPCODE_DEF: break;
+ case Shader::OPCODE_DCL: break;
+ case Shader::OPCODE_NOP: break;
+ case Shader::OPCODE_MOV: mov(d, s0); break;
+ case Shader::OPCODE_F2B: f2b(d, s0); break;
+ case Shader::OPCODE_B2F: b2f(d, s0); break;
+ case Shader::OPCODE_ADD: add(d, s0, s1); break;
+ case Shader::OPCODE_SUB: sub(d, s0, s1); break;
+ case Shader::OPCODE_MUL: mul(d, s0, s1); break;
+ case Shader::OPCODE_MAD: mad(d, s0, s1, s2); break;
+ case Shader::OPCODE_DP1: dp1(d, s0, s1); break;
+ case Shader::OPCODE_DP2: dp2(d, s0, s1); break;
+ case Shader::OPCODE_DP2ADD: dp2add(d, s0, s1, s2); break;
+ case Shader::OPCODE_DP3: dp3(d, s0, s1); break;
+ case Shader::OPCODE_DP4: dp4(d, s0, s1); break;
+ case Shader::OPCODE_CMP0: cmp0(d, s0, s1, s2); break;
+ case Shader::OPCODE_ICMP: icmp(d, s0, s1, control); break;
+ case Shader::OPCODE_SELECT: select(d, s0, s1, s2); break;
+ case Shader::OPCODE_EXTRACT: extract(d.x, s0, s1.x); break;
+ case Shader::OPCODE_INSERT: insert(d, s0, s1.x, s2.x); break;
+ case Shader::OPCODE_FRC: frc(d, s0); break;
+ case Shader::OPCODE_TRUNC: trunc(d, s0); break;
+ case Shader::OPCODE_FLOOR: floor(d, s0); break;
+ case Shader::OPCODE_CEIL: ceil(d, s0); break;
+ case Shader::OPCODE_EXP2X: exp2x(d, s0, pp); break;
+ case Shader::OPCODE_EXP2: exp2(d, s0, pp); break;
+ case Shader::OPCODE_LOG2X: log2x(d, s0, pp); break;
+ case Shader::OPCODE_LOG2: log2(d, s0, pp); break;
+ case Shader::OPCODE_EXP: exp(d, s0, pp); break;
+ case Shader::OPCODE_LOG: log(d, s0, pp); break;
+ case Shader::OPCODE_RCPX: rcpx(d, s0, pp); break;
+ case Shader::OPCODE_DIV: div(d, s0, s1); break;
+ case Shader::OPCODE_MOD: mod(d, s0, s1); break;
+ case Shader::OPCODE_RSQX: rsqx(d, s0, pp); break;
+ case Shader::OPCODE_SQRT: sqrt(d, s0, pp); break;
+ case Shader::OPCODE_RSQ: rsq(d, s0, pp); break;
+ case Shader::OPCODE_LEN2: len2(d.x, s0, pp); break;
+ case Shader::OPCODE_LEN3: len3(d.x, s0, pp); break;
+ case Shader::OPCODE_LEN4: len4(d.x, s0, pp); break;
+ case Shader::OPCODE_DIST1: dist1(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_DIST2: dist2(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_DIST3: dist3(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_DIST4: dist4(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_MIN: min(d, s0, s1); break;
+ case Shader::OPCODE_MAX: max(d, s0, s1); break;
+ case Shader::OPCODE_LRP: lrp(d, s0, s1, s2); break;
+ case Shader::OPCODE_STEP: step(d, s0, s1); break;
+ case Shader::OPCODE_SMOOTH: smooth(d, s0, s1, s2); break;
+ case Shader::OPCODE_POWX: powx(d, s0, s1, pp); break;
+ case Shader::OPCODE_POW: pow(d, s0, s1, pp); break;
+ case Shader::OPCODE_SGN: sgn(d, s0); break;
+ case Shader::OPCODE_CRS: crs(d, s0, s1); break;
+ case Shader::OPCODE_FORWARD1: forward1(d, s0, s1, s2); break;
+ case Shader::OPCODE_FORWARD2: forward2(d, s0, s1, s2); break;
+ case Shader::OPCODE_FORWARD3: forward3(d, s0, s1, s2); break;
+ case Shader::OPCODE_FORWARD4: forward4(d, s0, s1, s2); break;
+ case Shader::OPCODE_REFLECT1: reflect1(d, s0, s1); break;
+ case Shader::OPCODE_REFLECT2: reflect2(d, s0, s1); break;
+ case Shader::OPCODE_REFLECT3: reflect3(d, s0, s1); break;
+ case Shader::OPCODE_REFLECT4: reflect4(d, s0, s1); break;
+ case Shader::OPCODE_REFRACT1: refract1(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_REFRACT2: refract2(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_REFRACT3: refract3(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_REFRACT4: refract4(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_NRM2: nrm2(d, s0, pp); break;
+ case Shader::OPCODE_NRM3: nrm3(d, s0, pp); break;
+ case Shader::OPCODE_NRM4: nrm4(d, s0, pp); break;
+ case Shader::OPCODE_ABS: abs(d, s0); break;
+ case Shader::OPCODE_SINCOS: sincos(d, s0, pp); break;
+ case Shader::OPCODE_COS: cos(d, s0, pp); break;
+ case Shader::OPCODE_SIN: sin(d, s0, pp); break;
+ case Shader::OPCODE_TAN: tan(d, s0, pp); break;
+ case Shader::OPCODE_ACOS: acos(d, s0, pp); break;
+ case Shader::OPCODE_ASIN: asin(d, s0, pp); break;
+ case Shader::OPCODE_ATAN: atan(d, s0, pp); break;
+ case Shader::OPCODE_ATAN2: atan2(d, s0, s1, 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_TEXKILL: TEXKILL(cMask, d, dst.mask); break;
+ case Shader::OPCODE_DISCARD: DISCARD(r, 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_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_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_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;
+ case Shader::OPCODE_NOT: not(d, s0); break;
+ case Shader::OPCODE_OR: or(d.x, s0.x, s1.x); break;
+ case Shader::OPCODE_XOR: xor(d.x, s0.x, s1.x); break;
+ case Shader::OPCODE_AND: and(d.x, s0.x, s1.x); break;
+ case Shader::OPCODE_END: break;
default:
ASSERT(false);
}
- if(dst.type != Dst::PARAMETER_VOID && dst.type != Dst::PARAMETER_LABEL && opcode != Op::OPCODE_TEXKILL)
+ if(dst.type != Shader::PARAMETER_VOID && dst.type != Shader::PARAMETER_LABEL && opcode != Shader::OPCODE_TEXKILL && opcode != Shader::OPCODE_NOP)
{
- if(dst.saturate)
+ if(dst.integer)
{
- if(dst.x) d.r = Max(d.r, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- if(dst.y) d.g = Max(d.g, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- if(dst.z) d.b = Max(d.b, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- if(dst.w) d.a = Max(d.a, Float4(0.0f, 0.0f, 0.0f, 0.0f));
-
- if(dst.x) d.r = Min(d.r, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- if(dst.y) d.g = Min(d.g, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- if(dst.z) d.b = Min(d.b, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- if(dst.w) d.a = Min(d.a, Float4(1.0f, 1.0f, 1.0f, 1.0f));
+ switch(opcode)
+ {
+ case Shader::OPCODE_DIV:
+ if(dst.x) d.x = Trunc(d.x);
+ if(dst.y) d.y = Trunc(d.y);
+ if(dst.z) d.z = Trunc(d.z);
+ if(dst.w) d.w = Trunc(d.w);
+ break;
+ default:
+ break; // No truncation to integer required when arguments are integer
+ }
}
- if(pixelShader->containsDynamicBranching())
+ if(dst.saturate)
{
- Color4f pDst; // FIXME: Rename
+ if(dst.x) d.x = Max(d.x, Float4(0.0f));
+ if(dst.y) d.y = Max(d.y, Float4(0.0f));
+ if(dst.z) d.z = Max(d.z, Float4(0.0f));
+ if(dst.w) d.w = Max(d.w, Float4(0.0f));
+
+ if(dst.x) d.x = Min(d.x, Float4(1.0f));
+ if(dst.y) d.y = Min(d.y, Float4(1.0f));
+ if(dst.z) d.z = Min(d.z, Float4(1.0f));
+ if(dst.w) d.w = Min(d.w, Float4(1.0f));
+ }
+
+ if(shader->containsDynamicBranching())
+ {
+ Vector4f pDst; // FIXME: Rename
switch(dst.type)
{
- case Dst::PARAMETER_TEMP:
- if(dst.x) pDst.x = r.rf[dst.index].x;
- if(dst.y) pDst.y = r.rf[dst.index].y;
- if(dst.z) pDst.z = r.rf[dst.index].z;
- if(dst.w) pDst.w = r.rf[dst.index].w;
+ case Shader::PARAMETER_TEMP:
+ if(dst.rel.type == Shader::PARAMETER_VOID)
+ {
+ if(dst.x) pDst.x = r.rf[dst.index].x;
+ if(dst.y) pDst.y = r.rf[dst.index].y;
+ if(dst.z) pDst.z = r.rf[dst.index].z;
+ if(dst.w) pDst.w = r.rf[dst.index].w;
+ }
+ else
+ {
+ Int a = relativeAddress(r, dst);
+
+ if(dst.x) pDst.x = r.rf[dst.index + a].x;
+ if(dst.y) pDst.y = r.rf[dst.index + a].y;
+ if(dst.z) pDst.z = r.rf[dst.index + a].z;
+ if(dst.w) pDst.w = r.rf[dst.index + a].w;
+ }
break;
- case Dst::PARAMETER_COLOROUT:
+ case Shader::PARAMETER_COLOROUT:
+ ASSERT(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;
break;
- case Dst::PARAMETER_PREDICATE:
+ 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;
break;
- case Dst::PARAMETER_DEPTHOUT:
+ case Shader::PARAMETER_DEPTHOUT:
pDst.x = r.oDepth;
break;
default:
ASSERT(false);
}
- Int4 enable = r.enableStack[r.enableIndex] & r.enableBreak;
+ Int4 enable = enableMask(r, instruction);
Int4 xEnable = enable;
Int4 yEnable = enable;
@@ -3870,14 +4010,14 @@
if(predicate)
{
- unsigned char pSwizzle = instruction->getPredicateSwizzle();
+ 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];
- if(!instruction->isPredicateNot())
+ if(!instruction->predicateNot)
{
if(dst.x) xEnable = xEnable & As<Int4>(xPredicate);
if(dst.y) yEnable = yEnable & As<Int4>(yPredicate);
@@ -3906,25 +4046,38 @@
switch(dst.type)
{
- case Dst::PARAMETER_TEMP:
- if(dst.x) r.rf[dst.index].x = d.x;
- if(dst.y) r.rf[dst.index].y = d.y;
- if(dst.z) r.rf[dst.index].z = d.z;
- if(dst.w) r.rf[dst.index].w = d.w;
+ case Shader::PARAMETER_TEMP:
+ if(dst.rel.type == Shader::PARAMETER_VOID)
+ {
+ if(dst.x) r.rf[dst.index].x = d.x;
+ if(dst.y) r.rf[dst.index].y = d.y;
+ if(dst.z) r.rf[dst.index].z = d.z;
+ if(dst.w) r.rf[dst.index].w = d.w;
+ }
+ else
+ {
+ Int a = relativeAddress(r, dst);
+
+ if(dst.x) r.rf[dst.index + a].x = d.x;
+ if(dst.y) r.rf[dst.index + a].y = d.y;
+ if(dst.z) r.rf[dst.index + a].z = d.z;
+ if(dst.w) r.rf[dst.index + a].w = d.w;
+ }
break;
- case Dst::PARAMETER_COLOROUT:
- 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;
+ case Shader::PARAMETER_COLOROUT:
+ ASSERT(dst.rel.type == Shader::PARAMETER_VOID);
+ if(dst.x) {r.oC[dst.index].x = d.x; out[dst.index][0] = true;}
+ if(dst.y) {r.oC[dst.index].y = d.y; out[dst.index][1] = true;}
+ if(dst.z) {r.oC[dst.index].z = d.z; out[dst.index][2] = true;}
+ if(dst.w) {r.oC[dst.index].w = d.w; out[dst.index][3] = true;}
break;
- case Dst::PARAMETER_PREDICATE:
+ 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;
break;
- case Dst::PARAMETER_DEPTHOUT:
+ case Shader::PARAMETER_DEPTHOUT:
r.oDepth = d.x;
break;
default:
@@ -3933,36 +4086,47 @@
}
}
- if(returns)
+ if(currentLabel != -1)
{
Nucleus::setInsertBlock(returnBlock);
}
+
+ for(int i = 0; i < 4; i++)
+ {
+ if((Format)state.targetFormat[i] != FORMAT_NULL)
+ {
+ if(!out[i][0]) r.oC[i].x = Float4(0.0f);
+ if(!out[i][1]) r.oC[i].y = Float4(0.0f);
+ if(!out[i][2]) r.oC[i].z = Float4(0.0f);
+ if(!out[i][3]) r.oC[i].w = Float4(0.0f);
+ }
+ }
}
- Short4 PixelRoutine::convertFixed12(Float4 &cf)
+ Short4 PixelRoutine::convertFixed12(RValue<Float4> cf)
{
- return RoundShort4(cf * Float4(0x1000, 0x1000, 0x1000, 0x1000));
+ return RoundShort4(cf * Float4(0x1000));
}
- void PixelRoutine::convertFixed12(Color4i &ci, Color4f &cf)
+ void PixelRoutine::convertFixed12(Vector4i &ci, Vector4f &cf)
{
- ci.r = convertFixed12(cf.r);
- ci.g = convertFixed12(cf.g);
- ci.b = convertFixed12(cf.b);
- ci.a = convertFixed12(cf.a);
+ ci.x = convertFixed12(cf.x);
+ ci.y = convertFixed12(cf.y);
+ ci.z = convertFixed12(cf.z);
+ ci.w = convertFixed12(cf.w);
}
UShort4 PixelRoutine::convertFixed16(Float4 &cf, bool saturate)
{
- return UShort4(cf * Float4(0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF), saturate);
+ return UShort4(cf * Float4(0xFFFF), saturate);
}
- void PixelRoutine::convertFixed16(Color4i &ci, Color4f &cf, bool saturate)
+ void PixelRoutine::convertFixed16(Vector4i &ci, Vector4f &cf, bool saturate)
{
- ci.r = convertFixed16(cf.r, saturate);
- ci.g = convertFixed16(cf.g, saturate);
- ci.b = convertFixed16(cf.b, saturate);
- ci.a = convertFixed16(cf.a, saturate);
+ ci.x = convertFixed16(cf.x, saturate);
+ ci.y = convertFixed16(cf.y, saturate);
+ ci.z = convertFixed16(cf.z, saturate);
+ ci.w = convertFixed16(cf.w, saturate);
}
Float4 PixelRoutine::convertSigned12(Short4 &ci)
@@ -3970,75 +4134,75 @@
return Float4(ci) * Float4(1.0f / 0x0FFE);
}
- void PixelRoutine::convertSigned12(Color4f &cf, Color4i &ci)
+ void PixelRoutine::convertSigned12(Vector4f &cf, Vector4i &ci)
{
- cf.r = convertSigned12(ci.r);
- cf.g = convertSigned12(ci.g);
- cf.b = convertSigned12(ci.b);
- cf.a = convertSigned12(ci.a);
+ cf.x = convertSigned12(ci.x);
+ cf.y = convertSigned12(ci.y);
+ cf.z = convertSigned12(ci.z);
+ cf.w = convertSigned12(ci.w);
}
Float4 PixelRoutine::convertUnsigned16(UShort4 ci)
{
- return Float4(ci) * Float4(1.0f / 0xFFFF, 1.0f / 0xFFFF, 1.0f / 0xFFFF, 1.0f / 0xFFFF);
+ return Float4(ci) * Float4(1.0f / 0xFFFF);
}
- void PixelRoutine::sRGBtoLinear16_16(Registers &r, Color4i &c)
+ void PixelRoutine::sRGBtoLinear16_16(Registers &r, Vector4i &c)
{
- c.r = As<UShort4>(c.r) >> 4;
- c.g = As<UShort4>(c.g) >> 4;
- c.b = As<UShort4>(c.b) >> 4;
+ 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);
}
- void PixelRoutine::sRGBtoLinear12_16(Registers &r, Color4i &c)
+ void PixelRoutine::sRGBtoLinear12_16(Registers &r, Vector4i &c)
{
Pointer<Byte> LUT = r.constants + OFFSET(Constants,sRGBtoLin12_16);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 0))), 0);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 1))), 1);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 2))), 2);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 3))), 3);
+ 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);
+ c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 2))), 2);
+ c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 3))), 3);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 0))), 0);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 1))), 1);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 2))), 2);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 3))), 3);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 0))), 0);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 1))), 1);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 2))), 2);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 3))), 3);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 0))), 0);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 1))), 1);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 2))), 2);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 3))), 3);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 0))), 0);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 1))), 1);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 2))), 2);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 3))), 3);
}
- void PixelRoutine::linearToSRGB16_16(Registers &r, Color4i &c)
+ void PixelRoutine::linearToSRGB16_16(Registers &r, Vector4i &c)
{
- c.r = As<UShort4>(c.r) >> 4;
- c.g = As<UShort4>(c.g) >> 4;
- c.b = As<UShort4>(c.b) >> 4;
+ 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);
}
- void PixelRoutine::linearToSRGB12_16(Registers &r, Color4i &c)
+ void PixelRoutine::linearToSRGB12_16(Registers &r, Vector4i &c)
{
Pointer<Byte> LUT = r.constants + OFFSET(Constants,linToSRGB12_16);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 0))), 0);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 1))), 1);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 2))), 2);
- c.r = Insert(c.r, *Pointer<Short>(LUT + 2 * Int(Extract(c.r, 3))), 3);
+ 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);
+ c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 2))), 2);
+ c.x = Insert(c.x, *Pointer<Short>(LUT + 2 * Int(Extract(c.x, 3))), 3);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 0))), 0);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 1))), 1);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 2))), 2);
- c.g = Insert(c.g, *Pointer<Short>(LUT + 2 * Int(Extract(c.g, 3))), 3);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 0))), 0);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 1))), 1);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 2))), 2);
+ c.y = Insert(c.y, *Pointer<Short>(LUT + 2 * Int(Extract(c.y, 3))), 3);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 0))), 0);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 1))), 1);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 2))), 2);
- c.b = Insert(c.b, *Pointer<Short>(LUT + 2 * Int(Extract(c.b, 3))), 3);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 0))), 0);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 1))), 1);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 2))), 2);
+ c.z = Insert(c.z, *Pointer<Short>(LUT + 2 * Int(Extract(c.z, 3))), 3);
}
Float4 PixelRoutine::linearToSRGB(const Float4 &x) // Approximates x^(1.0/2.2)
@@ -4057,31 +4221,31 @@
return Min(Max(linear, Float4(0.0f)), Float4(1.0f));
}
- void PixelRoutine::MOV(Color4i &dst, Color4i &src0)
+ void PixelRoutine::MOV(Vector4i &dst, Vector4i &src0)
{
- dst.r = src0.x;
- dst.g = src0.y;
- dst.b = src0.z;
- dst.a = src0.w;
+ dst.x = src0.x;
+ dst.y = src0.y;
+ dst.z = src0.z;
+ dst.w = src0.w;
}
- void PixelRoutine::ADD(Color4i &dst, Color4i &src0, Color4i &src1)
+ void PixelRoutine::ADD(Vector4i &dst, Vector4i &src0, Vector4i &src1)
{
- dst.r = AddSat(src0.x, src1.x);
- dst.g = AddSat(src0.y, src1.y);
- dst.b = AddSat(src0.z, src1.z);
- dst.a = AddSat(src0.w, src1.w);
+ dst.x = AddSat(src0.x, src1.x);
+ dst.y = AddSat(src0.y, src1.y);
+ dst.z = AddSat(src0.z, src1.z);
+ dst.w = AddSat(src0.w, src1.w);
}
- void PixelRoutine::SUB(Color4i &dst, Color4i &src0, Color4i &src1)
+ void PixelRoutine::SUB(Vector4i &dst, Vector4i &src0, Vector4i &src1)
{
- dst.r = SubSat(src0.x, src1.x);
- dst.g = SubSat(src0.y, src1.y);
- dst.b = SubSat(src0.z, src1.z);
- dst.a = SubSat(src0.w, src1.w);
+ dst.x = SubSat(src0.x, src1.x);
+ dst.y = SubSat(src0.y, src1.y);
+ dst.z = SubSat(src0.z, src1.z);
+ dst.w = SubSat(src0.w, src1.w);
}
- void PixelRoutine::MAD(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2)
+ void PixelRoutine::MAD(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2)
{
// FIXME: Long fixed-point multiply fixup
{dst.x = MulHigh(src0.x, src1.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, src2.x);}
@@ -4090,7 +4254,7 @@
{dst.w = MulHigh(src0.w, src1.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, src2.w);}
}
- void PixelRoutine::MUL(Color4i &dst, Color4i &src0, Color4i &src1)
+ void PixelRoutine::MUL(Vector4i &dst, Vector4i &src0, Vector4i &src1)
{
// FIXME: Long fixed-point multiply fixup
{dst.x = MulHigh(src0.x, src1.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x);}
@@ -4099,7 +4263,7 @@
{dst.w = MulHigh(src0.w, src1.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w);}
}
- void PixelRoutine::DP3(Color4i &dst, Color4i &src0, Color4i &src1)
+ void PixelRoutine::DP3(Vector4i &dst, Vector4i &src0, Vector4i &src1)
{
Short4 t0;
Short4 t1;
@@ -4111,13 +4275,13 @@
t1 = MulHigh(src0.z, src1.z); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1);
t0 = AddSat(t0, t1);
- dst.r = t0;
- dst.g = t0;
- dst.b = t0;
- dst.a = t0;
+ dst.x = t0;
+ dst.y = t0;
+ dst.z = t0;
+ dst.w = t0;
}
- void PixelRoutine::DP4(Color4i &dst, Color4i &src0, Color4i &src1)
+ void PixelRoutine::DP4(Vector4i &dst, Vector4i &src0, Vector4i &src1)
{
Short4 t0;
Short4 t1;
@@ -4131,13 +4295,13 @@
t1 = MulHigh(src0.w, src1.w); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1);
t0 = AddSat(t0, t1);
- dst.r = t0;
- dst.g = t0;
- dst.b = t0;
- dst.a = t0;
+ dst.x = t0;
+ dst.y = t0;
+ dst.z = t0;
+ dst.w = t0;
}
- void PixelRoutine::LRP(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2)
+ void PixelRoutine::LRP(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2)
{
// FIXME: Long fixed-point multiply fixup
{dst.x = SubSat(src1.x, src2.x); dst.x = MulHigh(dst.x, src0.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, src2.x);}
@@ -4146,7 +4310,7 @@
{dst.w = SubSat(src1.w, src2.w); dst.w = MulHigh(dst.w, src0.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, src2.w);}
}
- void PixelRoutine::TEXCOORD(Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate)
+ void PixelRoutine::TEXCOORD(Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate)
{
Float4 uw;
Float4 vw;
@@ -4154,41 +4318,41 @@
if(state.interpolant[2 + coordinate].component & 0x01)
{
- uw = Max(u, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- uw = Min(uw, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- dst.r = convertFixed12(uw);
+ uw = Max(u, Float4(0.0f));
+ uw = Min(uw, Float4(1.0f));
+ dst.x = convertFixed12(uw);
}
else
{
- dst.r = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ dst.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
}
if(state.interpolant[2 + coordinate].component & 0x02)
{
- vw = Max(v, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- vw = Min(vw, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- dst.g = convertFixed12(vw);
+ vw = Max(v, Float4(0.0f));
+ vw = Min(vw, Float4(1.0f));
+ dst.y = convertFixed12(vw);
}
else
{
- dst.g = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ dst.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
}
if(state.interpolant[2 + coordinate].component & 0x04)
{
- sw = Max(s, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- sw = Min(sw, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- dst.b = convertFixed12(sw);
+ sw = Max(s, Float4(0.0f));
+ sw = Min(sw, Float4(1.0f));
+ dst.z = convertFixed12(sw);
}
else
{
- dst.b = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ dst.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
}
- dst.a = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ dst.w = Short4(0x1000);
}
- void PixelRoutine::TEXCRD(Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate, bool project)
+ void PixelRoutine::TEXCRD(Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate, bool project)
{
Float4 uw = u;
Float4 vw = v;
@@ -4202,68 +4366,68 @@
if(state.interpolant[2 + coordinate].component & 0x01)
{
- uw *= Float4(0x1000, 0x1000, 0x1000, 0x1000);
- uw = Max(uw, Float4(-0x8000, -0x8000, -0x8000, -0x8000));
- uw = Min(uw, Float4(0x7FFF, 0x7FFF, 0x7FFF, 0x7FFF));
- dst.r = RoundShort4(uw);
+ uw *= Float4(0x1000);
+ uw = Max(uw, Float4(-0x8000));
+ uw = Min(uw, Float4(0x7FFF));
+ dst.x = RoundShort4(uw);
}
else
{
- dst.r = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ dst.x = Short4(0x0000);
}
if(state.interpolant[2 + coordinate].component & 0x02)
{
- vw *= Float4(0x1000, 0x1000, 0x1000, 0x1000);
- vw = Max(vw, Float4(-0x8000, -0x8000, -0x8000, -0x8000));
- vw = Min(vw, Float4(0x7FFF, 0x7FFF, 0x7FFF, 0x7FFF));
- dst.g = RoundShort4(vw);
+ vw *= Float4(0x1000);
+ vw = Max(vw, Float4(-0x8000));
+ vw = Min(vw, Float4(0x7FFF));
+ dst.y = RoundShort4(vw);
}
else
{
- dst.g = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ dst.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
}
if(state.interpolant[2 + coordinate].component & 0x04)
{
- sw *= Float4(0x1000, 0x1000, 0x1000, 0x1000);
- sw = Max(sw, Float4(-0x8000, -0x8000, -0x8000, -0x8000));
- sw = Min(sw, Float4(0x7FFF, 0x7FFF, 0x7FFF, 0x7FFF));
- dst.b = RoundShort4(sw);
+ sw *= Float4(0x1000);
+ sw = Max(sw, Float4(-0x8000));
+ sw = Min(sw, Float4(0x7FFF));
+ dst.z = RoundShort4(sw);
}
else
{
- dst.b = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ dst.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
}
}
- void PixelRoutine::TEXDP3(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, Color4i &src)
+ void PixelRoutine::TEXDP3(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, Vector4i &src)
{
TEXM3X3PAD(r, u, v, s, src, 0, false);
- Short4 t0 = RoundShort4(r.u_ * Float4(0x1000, 0x1000, 0x1000, 0x1000));
+ Short4 t0 = RoundShort4(r.u_ * Float4(0x1000));
- dst.r = t0;
- dst.g = t0;
- dst.b = t0;
- dst.a = t0;
+ dst.x = t0;
+ dst.y = t0;
+ dst.z = t0;
+ dst.w = t0;
}
- void PixelRoutine::TEXDP3TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0)
+ void PixelRoutine::TEXDP3TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0)
{
TEXM3X3PAD(r, u, v, s, src0, 0, false);
- r.v_ = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.w_ = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ r.v_ = Float4(0.0f);
+ r.w_ = Float4(0.0f);
sampleTexture(r, dst, stage, r.u_, r.v_, r.w_, r.w_);
}
void PixelRoutine::TEXKILL(Int cMask[4], Float4 &u, Float4 &v, Float4 &s)
{
- Int kill = SignMask(CmpNLT(u, Float4(0, 0, 0, 0))) &
- SignMask(CmpNLT(v, Float4(0, 0, 0, 0))) &
- SignMask(CmpNLT(s, Float4(0, 0, 0, 0)));
+ Int kill = SignMask(CmpNLT(u, Float4(0.0f))) &
+ SignMask(CmpNLT(v, Float4(0.0f))) &
+ SignMask(CmpNLT(s, Float4(0.0f)));
for(unsigned int q = 0; q < state.multiSample; q++)
{
@@ -4271,9 +4435,9 @@
}
}
- void PixelRoutine::TEXKILL(Int cMask[4], Color4i &src)
+ void PixelRoutine::TEXKILL(Int cMask[4], Vector4i &src)
{
- Short4 test = src.r | src.g | src.b;
+ Short4 test = src.x | src.y | src.z;
Int kill = SignMask(Pack(test, test)) ^ 0x0000000F;
for(unsigned int q = 0; q < state.multiSample; q++)
@@ -4282,24 +4446,24 @@
}
}
- void PixelRoutine::TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int sampler, bool project)
+ void PixelRoutine::TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int sampler, bool project)
{
sampleTexture(r, dst, sampler, u, v, s, s, project);
}
- void PixelRoutine::TEXLD(Registers &r, Color4i &dst, Color4i &src, int sampler, bool project)
+ void PixelRoutine::TEXLD(Registers &r, Vector4i &dst, Vector4i &src, int sampler, bool project)
{
- Float4 u = Float4(src.r) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 v = Float4(src.g) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 s = Float4(src.b) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
+ 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);
}
- void PixelRoutine::TEXBEM(Registers &r, Color4i &dst, Color4i &src, Float4 &u, Float4 &v, Float4 &s, int stage)
+ void PixelRoutine::TEXBEM(Registers &r, Vector4i &dst, Vector4i &src, Float4 &u, Float4 &v, Float4 &s, int stage)
{
- Float4 du = Float4(src.r) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 dv = Float4(src.g) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
+ Float4 du = Float4(src.x) * Float4(1.0f / 0x0FFE);
+ Float4 dv = Float4(src.y) * Float4(1.0f / 0x0FFE);
Float4 du2 = du;
Float4 dv2 = dv;
@@ -4317,10 +4481,10 @@
sampleTexture(r, dst, stage, u_, v_, s, s);
}
- void PixelRoutine::TEXBEML(Registers &r, Color4i &dst, Color4i &src, Float4 &u, Float4 &v, Float4 &s, int stage)
+ void PixelRoutine::TEXBEML(Registers &r, Vector4i &dst, Vector4i &src, Float4 &u, Float4 &v, Float4 &s, int stage)
{
- Float4 du = Float4(src.r) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 dv = Float4(src.g) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
+ Float4 du = Float4(src.x) * Float4(1.0f / 0x0FFE);
+ Float4 dv = Float4(src.y) * Float4(1.0f / 0x0FFE);
Float4 du2 = du;
Float4 dv2 = dv;
@@ -4339,46 +4503,46 @@
Short4 L;
- L = src.b;
+ L = src.z;
L = MulHigh(L, *Pointer<Short4>(r.data + OFFSET(DrawData,textureStage[stage].luminanceScale4)));
L = L << 4;
L = AddSat(L, *Pointer<Short4>(r.data + OFFSET(DrawData,textureStage[stage].luminanceOffset4)));
L = Max(L, Short4(0x0000, 0x0000, 0x0000, 0x0000));
- L = Min(L, Short4(0x1000, 0x1000, 0x1000, 0x1000));
+ L = Min(L, Short4(0x1000));
- dst.r = MulHigh(dst.r, L); dst.r = dst.r << 4;
- dst.g = MulHigh(dst.g, L); dst.g = dst.g << 4;
- dst.b = MulHigh(dst.b, L); dst.b = dst.b << 4;
+ dst.x = MulHigh(dst.x, L); dst.x = dst.x << 4;
+ dst.y = MulHigh(dst.y, L); dst.y = dst.y << 4;
+ dst.z = MulHigh(dst.z, L); dst.z = dst.z << 4;
}
- void PixelRoutine::TEXREG2AR(Registers &r, Color4i &dst, Color4i &src0, int stage)
+ void PixelRoutine::TEXREG2AR(Registers &r, Vector4i &dst, Vector4i &src0, int stage)
{
- Float4 u = Float4(src0.a) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 v = Float4(src0.r) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 s = Float4(src0.b) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
+ 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);
}
- void PixelRoutine::TEXREG2GB(Registers &r, Color4i &dst, Color4i &src0, int stage)
+ void PixelRoutine::TEXREG2GB(Registers &r, Vector4i &dst, Vector4i &src0, int stage)
{
- Float4 u = Float4(src0.g) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 v = Float4(src0.b) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
+ 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);
}
- void PixelRoutine::TEXREG2RGB(Registers &r, Color4i &dst, Color4i &src0, int stage)
+ void PixelRoutine::TEXREG2RGB(Registers &r, Vector4i &dst, Vector4i &src0, int stage)
{
- Float4 u = Float4(src0.r) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 v = Float4(src0.g) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- Float4 s = Float4(src0.b) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
+ 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);
}
- void PixelRoutine::TEXM3X2DEPTH(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, Color4i &src, bool signedScaling)
+ void PixelRoutine::TEXM3X2DEPTH(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, Vector4i &src, bool signedScaling)
{
TEXM3X2PAD(r, u, v, s, src, 1, signedScaling);
@@ -4388,44 +4552,44 @@
r.oDepth = r.u_;
}
- void PixelRoutine::TEXM3X2PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Color4i &src0, int component, bool signedScaling)
+ void PixelRoutine::TEXM3X2PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4i &src0, int component, bool signedScaling)
{
TEXM3X3PAD(r, u, v, s, src0, component, signedScaling);
}
- void PixelRoutine::TEXM3X2TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0, bool signedScaling)
+ void PixelRoutine::TEXM3X2TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0, bool signedScaling)
{
TEXM3X2PAD(r, u, v, s, src0, 1, signedScaling);
- r.w_ = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ r.w_ = Float4(0.0f);
sampleTexture(r, dst, stage, r.u_, r.v_, r.w_, r.w_);
}
- void PixelRoutine::TEXM3X3(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, Color4i &src0, bool signedScaling)
+ void PixelRoutine::TEXM3X3(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, Vector4i &src0, bool signedScaling)
{
TEXM3X3PAD(r, u, v, s, src0, 2, signedScaling);
- dst.r = RoundShort4(r.u_ * Float4(0x1000, 0x1000, 0x1000, 0x1000));
- dst.g = RoundShort4(r.v_ * Float4(0x1000, 0x1000, 0x1000, 0x1000));
- dst.b = RoundShort4(r.w_ * Float4(0x1000, 0x1000, 0x1000, 0x1000));
- dst.a = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ dst.x = RoundShort4(r.u_ * Float4(0x1000));
+ dst.y = RoundShort4(r.v_ * Float4(0x1000));
+ dst.z = RoundShort4(r.w_ * Float4(0x1000));
+ dst.w = Short4(0x1000);
}
- void PixelRoutine::TEXM3X3PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Color4i &src0, int component, bool signedScaling)
+ void PixelRoutine::TEXM3X3PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4i &src0, int component, bool signedScaling)
{
if(component == 0 || previousScaling != signedScaling) // FIXME: Other source modifiers?
{
- r.U = Float4(src0.r);
- r.V = Float4(src0.g);
- r.W = Float4(src0.b);
+ r.U = Float4(src0.x);
+ r.V = Float4(src0.y);
+ r.W = Float4(src0.z);
previousScaling = signedScaling;
}
Float4 x = r.U * u + r.V * v + r.W * s;
- x *= Float4(1.0f / 0x1000, 1.0f / 0x1000, 1.0f / 0x1000, 1.0f / 0x1000);
+ x *= Float4(1.0f / 0x1000);
switch(component)
{
@@ -4436,15 +4600,15 @@
}
}
- void PixelRoutine::TEXM3X3SPEC(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0, Color4i &src1)
+ void PixelRoutine::TEXM3X3SPEC(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0, Vector4i &src1)
{
TEXM3X3PAD(r, u, v, s, src0, 2, false);
Float4 E[3]; // Eye vector
- E[0] = Float4(src1.r) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- E[1] = Float4(src1.g) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
- E[2] = Float4(src1.b) * Float4(1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE, 1.0f / 0x0FFE);
+ E[0] = Float4(src1.x) * Float4(1.0f / 0x0FFE);
+ E[1] = Float4(src1.y) * Float4(1.0f / 0x0FFE);
+ E[2] = Float4(src1.z) * Float4(1.0f / 0x0FFE);
// Reflection
Float4 u__;
@@ -4473,22 +4637,22 @@
sampleTexture(r, dst, stage, u__, v__, w__, w__);
}
- void PixelRoutine::TEXM3X3TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0, bool signedScaling)
+ void PixelRoutine::TEXM3X3TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0, bool signedScaling)
{
TEXM3X3PAD(r, u, v, s, src0, 2, signedScaling);
sampleTexture(r, dst, stage, r.u_, r.v_, r.w_, r.w_);
}
- void PixelRoutine::TEXM3X3VSPEC(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0)
+ void PixelRoutine::TEXM3X3VSPEC(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0)
{
TEXM3X3PAD(r, u, v, s, src0, 2, false);
Float4 E[3]; // Eye vector
- E[0] = r.vw[2 + stage - 2];
- E[1] = r.vw[2 + stage - 1];
- E[2] = r.vw[2 + stage - 0];
+ E[0] = r.vf[2 + stage - 2].w;
+ E[1] = r.vf[2 + stage - 1].w;
+ E[2] = r.vf[2 + stage - 0].w;
// Reflection
Float4 u__;
@@ -4519,8 +4683,8 @@
void PixelRoutine::TEXDEPTH(Registers &r)
{
- r.u_ = Float4(r.ri[5].r);
- r.v_ = Float4(r.ri[5].g);
+ r.u_ = Float4(r.ri[5].x);
+ r.v_ = Float4(r.ri[5].y);
// z / w
r.u_ *= Rcp_pp(r.v_); // FIXME: Set result to 1.0 when division by zero
@@ -4528,68 +4692,68 @@
r.oDepth = r.u_;
}
- void PixelRoutine::CND(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2)
+ void PixelRoutine::CND(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2)
{
- {Short4 t0; t0 = src0.x; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.x; t1 = t1 & t0; t0 = ~t0 & src2.x; t0 = t0 | t1; dst.r = t0;};
- {Short4 t0; t0 = src0.y; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.y; t1 = t1 & t0; t0 = ~t0 & src2.y; t0 = t0 | t1; dst.g = t0;};
- {Short4 t0; t0 = src0.z; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.z; t1 = t1 & t0; t0 = ~t0 & src2.z; t0 = t0 | t1; dst.b = t0;};
- {Short4 t0; t0 = src0.w; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.w; t1 = t1 & t0; t0 = ~t0 & src2.w; t0 = t0 | t1; dst.a = t0;};
+ {Short4 t0; t0 = src0.x; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.x; t1 = t1 & t0; t0 = ~t0 & src2.x; t0 = t0 | t1; dst.x = t0;};
+ {Short4 t0; t0 = src0.y; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.y; t1 = t1 & t0; t0 = ~t0 & src2.y; t0 = t0 | t1; dst.y = t0;};
+ {Short4 t0; t0 = src0.z; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.z; t1 = t1 & t0; t0 = ~t0 & src2.z; t0 = t0 | t1; dst.z = t0;};
+ {Short4 t0; t0 = src0.w; t0 = CmpGT(t0, Short4(0x0800, 0x0800, 0x0800, 0x0800)); Short4 t1; t1 = src1.w; t1 = t1 & t0; t0 = ~t0 & src2.w; t0 = t0 | t1; dst.w = t0;};
}
- void PixelRoutine::CMP(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2)
+ void PixelRoutine::CMP(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2)
{
- {Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.x); Short4 t1; t1 = src2.x; t1 &= t0; t0 = ~t0 & src1.x; t0 |= t1; dst.r = t0;};
- {Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.y); Short4 t1; t1 = src2.y; t1 &= t0; t0 = ~t0 & src1.y; t0 |= t1; dst.g = t0;};
- {Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.z); Short4 t1; t1 = src2.z; t1 &= t0; t0 = ~t0 & src1.z; t0 |= t1; dst.b = t0;};
- {Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.w); Short4 t1; t1 = src2.w; t1 &= t0; t0 = ~t0 & src1.w; t0 |= t1; dst.a = t0;};
+ {Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.x); Short4 t1; t1 = src2.x; t1 &= t0; t0 = ~t0 & src1.x; t0 |= t1; dst.x = t0;};
+ {Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.y); Short4 t1; t1 = src2.y; t1 &= t0; t0 = ~t0 & src1.y; t0 |= t1; dst.y = t0;};
+ {Short4 t0 = CmpGT(Short4(0x0000, 0x0000, 0x0000, 0x0000), src0.z); Short4 t1; t1 = src2.z; t1 &= t0; t0 = ~t0 & src1.z; t0 |= t1; dst.z = t0;};
+ {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 PixelRoutine::BEM(Registers &r, Color4i &dst, Color4i &src0, Color4i &src1, int stage)
+ void PixelRoutine::BEM(Registers &r, Vector4i &dst, Vector4i &src0, Vector4i &src1, int stage)
{
Short4 t0;
Short4 t1;
- // dst.r = src0.r + BUMPENVMAT00(stage) * src1.r + BUMPENVMAT10(stage) * src1.g
+ // 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 = AddSat(t0, t1);
t0 = AddSat(t0, src0.x);
- dst.r = t0;
+ dst.x = t0;
- // dst.g = src0.g + BUMPENVMAT01(stage) * src1.r + BUMPENVMAT11(stage) * src1.g
+ // 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 = AddSat(t0, t1);
t0 = AddSat(t0, src0.y);
- dst.g = t0;
+ dst.y = t0;
}
- void PixelRoutine::M3X2(Registers &r, Color4f &dst, Color4f &src0, const Src &src1)
+ void PixelRoutine::M3X2(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
}
- void PixelRoutine::M3X3(Registers &r, Color4f &dst, Color4f &src0, const Src &src1)
+ void PixelRoutine::M3X3(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
dst.z = dot3(src0, row2);
}
- void PixelRoutine::M3X4(Registers &r, Color4f &dst, Color4f &src0, const Src &src1)
+ void PixelRoutine::M3X4(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
- Color4f row3 = reg(r, src1, 3);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
+ Vector4f row3 = reg(r, src1, 3);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
@@ -4597,23 +4761,23 @@
dst.w = dot3(src0, row3);
}
- void PixelRoutine::M4X3(Registers &r, Color4f &dst, Color4f &src0, const Src &src1)
+ void PixelRoutine::M4X3(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1);
dst.z = dot4(src0, row2);
}
- void PixelRoutine::M4X4(Registers &r, Color4f &dst, Color4f &src0, const Src &src1)
+ void PixelRoutine::M4X4(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
- Color4f row3 = reg(r, src1, 3);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
+ Vector4f row3 = reg(r, src1, 3);
dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1);
@@ -4621,11 +4785,10 @@
dst.w = dot4(src0, row3);
}
- void PixelRoutine::TEXLD(Registers &r, Color4f &dst, Color4f &src0, const Src &src1, bool project, bool bias)
+ void PixelRoutine::TEXLD(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias)
{
- Color4f tmp;
-
- sampleTexture(r, tmp, src1.index, src0.u, src0.v, src0.s, src0.t, src0, src0, project, bias);
+ Vector4f tmp;
+ sampleTexture(r, 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];
@@ -4633,11 +4796,10 @@
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
}
- void PixelRoutine::TEXLDD(Registers &r, Color4f &dst, Color4f &src0, const Src &src1, Color4f &src2, Color4f &src3, bool project, bool bias)
+ void PixelRoutine::TEXLDD(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, Vector4f &src2, Vector4f &src3, bool project, bool bias)
{
- Color4f tmp;
-
- sampleTexture(r, tmp, src1.index, src0.u, src0.v, src0.s, src0.t, src2, src3, project, bias, true);
+ Vector4f tmp;
+ sampleTexture(r, 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];
@@ -4645,11 +4807,10 @@
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
}
- void PixelRoutine::TEXLDL(Registers &r, Color4f &dst, Color4f &src0, const Src &src1, bool project, bool bias)
+ void PixelRoutine::TEXLDL(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1, bool project, bool bias)
{
- Color4f tmp;
-
- sampleTexture(r, tmp, src1.index, src0.u, src0.v, src0.s, src0.t, src0, src0, project, bias, false, true);
+ Vector4f tmp;
+ sampleTexture(r, 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];
@@ -4657,35 +4818,69 @@
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
}
- void PixelRoutine::TEXKILL(Int cMask[4], Color4f &src, unsigned char mask)
+ void PixelRoutine::TEXKILL(Int cMask[4], Vector4f &src, unsigned char mask)
{
Int kill = -1;
- if(mask & 0x1) kill &= SignMask(CmpNLT(src.x, Float4(0, 0, 0, 0)));
- if(mask & 0x2) kill &= SignMask(CmpNLT(src.y, Float4(0, 0, 0, 0)));
- if(mask & 0x4) kill &= SignMask(CmpNLT(src.z, Float4(0, 0, 0, 0)));
- if(mask & 0x8) kill &= SignMask(CmpNLT(src.w, Float4(0, 0, 0, 0)));
+ if(mask & 0x1) kill &= SignMask(CmpNLT(src.x, Float4(0.0f)));
+ if(mask & 0x2) kill &= SignMask(CmpNLT(src.y, Float4(0.0f)));
+ if(mask & 0x4) kill &= SignMask(CmpNLT(src.z, Float4(0.0f)));
+ if(mask & 0x8) kill &= SignMask(CmpNLT(src.w, Float4(0.0f)));
+
+ // FIXME: Dynamic branching affects TEXKILL?
+ // if(shader->containsDynamicBranching())
+ // {
+ // kill = ~SignMask(enableMask(r));
+ // }
for(unsigned int q = 0; q < state.multiSample; q++)
{
cMask[q] &= kill;
}
+
+ // FIXME: Branch to end of shader if all killed?
}
- void PixelRoutine::DSX(Color4f &dst, Color4f &src)
+ void PixelRoutine::DISCARD(Registers &r, Int cMask[4], const Shader::Instruction *instruction)
{
- dst.x = src.x.yyyy - src.x.xxxx;
- dst.y = src.y.yyyy - src.y.xxxx;
- dst.z = src.z.yyyy - src.z.xxxx;
- dst.w = src.w.yyyy - src.w.xxxx;
+ Int kill = 0;
+
+ if(shader->containsDynamicBranching())
+ {
+ kill = ~SignMask(enableMask(r, instruction));
+ }
+
+ for(unsigned int q = 0; q < state.multiSample; q++)
+ {
+ cMask[q] &= kill;
+ }
+
+ // FIXME: Branch to end of shader if all killed?
}
- void PixelRoutine::DSY(Color4f &dst, Color4f &src)
+ void PixelRoutine::DFDX(Vector4f &dst, Vector4f &src)
{
- dst.x = src.x.zzzz - src.x.xxxx;
- dst.y = src.y.zzzz - src.y.xxxx;
- dst.z = src.z.zzzz - src.z.xxxx;
- dst.w = src.w.zzzz - src.w.xxxx;
+ dst.x = src.x.yyww - src.x.xxzz;
+ dst.y = src.y.yyww - src.y.xxzz;
+ dst.z = src.z.yyww - src.z.xxzz;
+ dst.w = src.w.yyww - src.w.xxzz;
+ }
+
+ void PixelRoutine::DFDY(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = src.x.zwzw - src.x.xyxy;
+ dst.y = src.y.zwzw - src.y.xyxy;
+ dst.z = src.z.zwzw - src.z.xyxy;
+ dst.w = src.w.zwzw - src.w.xyxy;
+ }
+
+ void PixelRoutine::FWIDTH(Vector4f &dst, Vector4f &src)
+ {
+ // abs(dFdx(src)) + abs(dFdy(src));
+ dst.x = Abs(src.x.yyww - src.x.xxzz) + Abs(src.x.zwzw - src.x.xyxy);
+ dst.y = Abs(src.y.yyww - src.x.xxzz) + Abs(src.y.zwzw - src.y.xyxy);
+ dst.z = Abs(src.z.yyww - src.x.xxzz) + Abs(src.z.zwzw - src.z.xyxy);
+ dst.w = Abs(src.w.yyww - src.x.xxzz) + Abs(src.w.zwzw - src.w.xyxy);
}
void PixelRoutine::BREAK(Registers &r)
@@ -4695,6 +4890,7 @@
if(breakDepth == 0)
{
+ r.enableIndex = r.enableIndex - breakDepth;
Nucleus::createBr(endBlock);
}
else
@@ -4702,49 +4898,47 @@
r.enableBreak = r.enableBreak & ~r.enableStack[r.enableIndex];
Bool allBreak = SignMask(r.enableBreak) == 0x0;
+ r.enableIndex = r.enableIndex - breakDepth;
branch(allBreak, endBlock, deadBlock);
}
Nucleus::setInsertBlock(deadBlock);
+ r.enableIndex = r.enableIndex + breakDepth;
}
- void PixelRoutine::BREAKC(Registers &r, Color4f &src0, Color4f &src1, Control control)
+ void PixelRoutine::BREAKC(Registers &r, Vector4f &src0, Vector4f &src1, Control control)
{
Int4 condition;
switch(control)
{
- case Op::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
- case Op::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
- case Op::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
- case Op::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
- case Op::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
- case Op::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
+ case Shader::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
+ case Shader::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
+ case Shader::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
+ case Shader::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
default:
ASSERT(false);
}
- condition &= r.enableStack[r.enableIndex];
-
- llvm::BasicBlock *continueBlock = Nucleus::createBasicBlock();
- llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth - 1];
-
- r.enableBreak = r.enableBreak & ~condition;
- Bool allBreak = SignMask(r.enableBreak) == 0x0;
-
- branch(allBreak, endBlock, continueBlock);
- Nucleus::setInsertBlock(continueBlock);
+ BREAK(r, condition);
}
void PixelRoutine::BREAKP(Registers &r, const Src &predicateRegister) // FIXME: Factor out parts common with BREAKC
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
- if(predicateRegister.modifier == Src::MODIFIER_NOT)
+ if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
+ BREAK(r, condition);
+ }
+
+ void PixelRoutine::BREAK(Registers &r, Int4 &condition)
+ {
condition &= r.enableStack[r.enableIndex];
llvm::BasicBlock *continueBlock = Nucleus::createBasicBlock();
@@ -4753,44 +4947,61 @@
r.enableBreak = r.enableBreak & ~condition;
Bool allBreak = SignMask(r.enableBreak) == 0x0;
+ r.enableIndex = r.enableIndex - breakDepth;
branch(allBreak, endBlock, continueBlock);
+
Nucleus::setInsertBlock(continueBlock);
+ r.enableIndex = r.enableIndex + breakDepth;
}
- void PixelRoutine::CALL(Registers &r, int labelIndex)
+ void PixelRoutine::CONTINUE(Registers &r)
+ {
+ r.enableContinue = r.enableContinue & ~r.enableStack[r.enableIndex];
+ }
+
+ void PixelRoutine::TEST()
+ {
+ whileTest = true;
+ }
+
+ void PixelRoutine::CALL(Registers &r, int labelIndex, int callSiteIndex)
{
if(!labelBlock[labelIndex])
{
labelBlock[labelIndex] = Nucleus::createBasicBlock();
}
- llvm::BasicBlock *retBlock = Nucleus::createBasicBlock();
- callRetBlock.push_back(retBlock);
+ if(callRetBlock[labelIndex].size() > 1)
+ {
+ r.callStack[r.stackIndex++] = UInt(callSiteIndex);
+ }
- r.callStack[r.stackIndex++] = UInt((int)callRetBlock.size() - 1); // FIXME
+ Int4 restoreLeave = r.enableLeave;
Nucleus::createBr(labelBlock[labelIndex]);
- Nucleus::setInsertBlock(retBlock);
+ Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
+
+ r.enableLeave = restoreLeave;
}
- void PixelRoutine::CALLNZ(Registers &r, int labelIndex, const Src &src)
+ void PixelRoutine::CALLNZ(Registers &r, int labelIndex, int callSiteIndex, const Src &src)
{
- if(src.type == Src::PARAMETER_CONSTBOOL)
+ if(src.type == Shader::PARAMETER_CONSTBOOL)
{
- CALLNZb(r, labelIndex, src);
+ CALLNZb(r, labelIndex, callSiteIndex, src);
}
- else if(src.type == Src::PARAMETER_PREDICATE)
+ else if(src.type == Shader::PARAMETER_PREDICATE)
{
- CALLNZp(r, labelIndex, src);
+ CALLNZp(r, labelIndex, callSiteIndex, src);
}
else ASSERT(false);
}
- void PixelRoutine::CALLNZb(Registers &r, int labelIndex, const Src &boolRegister)
+ void PixelRoutine::CALLNZb(Registers &r, int labelIndex, int callSiteIndex, const Src &boolRegister)
{
Bool condition = (*Pointer<Byte>(r.data + OFFSET(DrawData,ps.b[boolRegister.index])) != Byte(0)); // FIXME
- if(boolRegister.modifier == Src::MODIFIER_NOT)
+ if(boolRegister.modifier == Shader::MODIFIER_NOT)
{
condition = !condition;
}
@@ -4800,20 +5011,24 @@
labelBlock[labelIndex] = Nucleus::createBasicBlock();
}
- llvm::BasicBlock *retBlock = Nucleus::createBasicBlock();
- callRetBlock.push_back(retBlock);
+ if(callRetBlock[labelIndex].size() > 1)
+ {
+ r.callStack[r.stackIndex++] = UInt(callSiteIndex);
+ }
- r.callStack[r.stackIndex++] = UInt((int)callRetBlock.size() - 1); // FIXME
+ Int4 restoreLeave = r.enableLeave;
- branch(condition, labelBlock[labelIndex], retBlock);
- Nucleus::setInsertBlock(retBlock);
+ branch(condition, labelBlock[labelIndex], callRetBlock[labelIndex][callSiteIndex]);
+ Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
+
+ r.enableLeave = restoreLeave;
}
- void PixelRoutine::CALLNZp(Registers &r, int labelIndex, const Src &predicateRegister)
+ void PixelRoutine::CALLNZp(Registers &r, int labelIndex, int callSiteIndex, const Src &predicateRegister)
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
- if(predicateRegister.modifier == Src::MODIFIER_NOT)
+ if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
@@ -4825,20 +5040,21 @@
labelBlock[labelIndex] = Nucleus::createBasicBlock();
}
- llvm::BasicBlock *retBlock = Nucleus::createBasicBlock();
- callRetBlock.push_back(retBlock);
-
- r.callStack[r.stackIndex++] = UInt((int)callRetBlock.size() - 1); // FIXME
+ if(callRetBlock[labelIndex].size() > 1)
+ {
+ r.callStack[r.stackIndex++] = UInt(callSiteIndex);
+ }
r.enableIndex++;
r.enableStack[r.enableIndex] = condition;
+ Int4 restoreLeave = r.enableLeave;
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
-
- branch(notAllFalse, labelBlock[labelIndex], retBlock);
- Nucleus::setInsertBlock(retBlock);
+ Bool notAllFalse = SignMask(condition) != 0;
+ branch(notAllFalse, labelBlock[labelIndex], callRetBlock[labelIndex][callSiteIndex]);
+ Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
r.enableIndex--;
+ r.enableLeave = restoreLeave;
}
void PixelRoutine::ELSE(Registers &r)
@@ -4851,7 +5067,7 @@
if(isConditionalIf[ifDepth])
{
Int4 condition = ~r.enableStack[r.enableIndex] & r.enableStack[r.enableIndex - 1];
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
+ Bool notAllFalse = SignMask(condition) != 0;
branch(notAllFalse, falseBlock, endBlock);
@@ -4884,20 +5100,6 @@
}
}
- void PixelRoutine::ENDREP(Registers &r)
- {
- loopRepDepth--;
-
- llvm::BasicBlock *testBlock = loopRepTestBlock[loopRepDepth];
- llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth];
-
- Nucleus::createBr(testBlock);
- Nucleus::setInsertBlock(endBlock);
-
- r.loopDepth--;
- r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
- }
-
void PixelRoutine::ENDLOOP(Registers &r)
{
loopRepDepth--;
@@ -4914,26 +5116,61 @@
r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
+ void PixelRoutine::ENDREP(Registers &r)
+ {
+ loopRepDepth--;
+
+ llvm::BasicBlock *testBlock = loopRepTestBlock[loopRepDepth];
+ llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth];
+
+ Nucleus::createBr(testBlock);
+ Nucleus::setInsertBlock(endBlock);
+
+ r.loopDepth--;
+ r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+ }
+
+ void PixelRoutine::ENDWHILE(Registers &r)
+ {
+ loopRepDepth--;
+
+ llvm::BasicBlock *testBlock = loopRepTestBlock[loopRepDepth];
+ llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth];
+
+ Nucleus::createBr(testBlock);
+ Nucleus::setInsertBlock(endBlock);
+
+ r.enableIndex--;
+ r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+ whileTest = false;
+ }
+
void PixelRoutine::IF(Registers &r, const Src &src)
{
- if(src.type == Src::PARAMETER_CONSTBOOL)
+ if(src.type == Shader::PARAMETER_CONSTBOOL)
{
IFb(r, src);
}
- else if(src.type == Src::PARAMETER_PREDICATE)
+ else if(src.type == Shader::PARAMETER_PREDICATE)
{
IFp(r, src);
}
- else ASSERT(false);
+ else
+ {
+ Int4 condition = As<Int4>(reg(r, src).x);
+ IF(r, condition);
+ }
}
void PixelRoutine::IFb(Registers &r, const Src &boolRegister)
{
+ ASSERT(ifDepth < 24 + 4);
+
Bool condition = (*Pointer<Byte>(r.data + OFFSET(DrawData,ps.b[boolRegister.index])) != Byte(0)); // FIXME
- if(boolRegister.modifier == Src::MODIFIER_NOT)
+ if(boolRegister.modifier == Shader::MODIFIER_NOT)
{
- condition = !condition;
+ condition = !condition;
}
llvm::BasicBlock *trueBlock = Nucleus::createBasicBlock();
@@ -4947,50 +5184,39 @@
ifDepth++;
}
- void PixelRoutine::IFp(Registers &r, const Src &predicateRegister) // FIXME: Factor out parts common with IFC
+ void PixelRoutine::IFp(Registers &r, const Src &predicateRegister)
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
- if(predicateRegister.modifier == Src::MODIFIER_NOT)
+ if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
- condition &= r.enableStack[r.enableIndex];
-
- r.enableIndex++;
- r.enableStack[r.enableIndex] = condition;
-
- llvm::BasicBlock *trueBlock = Nucleus::createBasicBlock();
- llvm::BasicBlock *falseBlock = Nucleus::createBasicBlock();
-
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
-
- branch(notAllFalse, trueBlock, falseBlock);
-
- isConditionalIf[ifDepth] = true;
- ifFalseBlock[ifDepth] = falseBlock;
-
- ifDepth++;
- breakDepth++;
+ IF(r, condition);
}
- void PixelRoutine::IFC(Registers &r, Color4f &src0, Color4f &src1, Control control)
+ void PixelRoutine::IFC(Registers &r, Vector4f &src0, Vector4f &src1, Control control)
{
Int4 condition;
switch(control)
{
- case Op::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
- case Op::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
- case Op::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
- case Op::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
- case Op::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
- case Op::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
+ case Shader::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
+ case Shader::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
+ case Shader::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
+ case Shader::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
default:
ASSERT(false);
}
+ IF(r, condition);
+ }
+
+ void PixelRoutine::IF(Registers &r, Int4 &condition)
+ {
condition &= r.enableStack[r.enableIndex];
r.enableIndex++;
@@ -4999,7 +5225,7 @@
llvm::BasicBlock *trueBlock = Nucleus::createBasicBlock();
llvm::BasicBlock *falseBlock = Nucleus::createBasicBlock();
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
+ Bool notAllFalse = SignMask(condition) != 0;
branch(notAllFalse, trueBlock, falseBlock);
@@ -5012,7 +5238,13 @@
void PixelRoutine::LABEL(int labelIndex)
{
+ if(!labelBlock[labelIndex])
+ {
+ labelBlock[labelIndex] = Nucleus::createBasicBlock();
+ }
+
Nucleus::setInsertBlock(labelBlock[labelIndex]);
+ currentLabel = labelIndex;
}
void PixelRoutine::LOOP(Registers &r, const Src &integerRegister)
@@ -5075,27 +5307,73 @@
breakDepth = 0;
}
+ void PixelRoutine::WHILE(Registers &r, const Src &temporaryRegister)
+ {
+ r.enableIndex++;
+
+ llvm::BasicBlock *loopBlock = Nucleus::createBasicBlock();
+ llvm::BasicBlock *testBlock = Nucleus::createBasicBlock();
+ llvm::BasicBlock *endBlock = Nucleus::createBasicBlock();
+
+ loopRepTestBlock[loopRepDepth] = testBlock;
+ loopRepEndBlock[loopRepDepth] = endBlock;
+
+ Int4 restoreBreak = r.enableBreak;
+ Int4 restoreContinue = r.enableContinue;
+
+ // FIXME: jump(testBlock)
+ Nucleus::createBr(testBlock);
+ Nucleus::setInsertBlock(testBlock);
+ r.enableContinue = restoreContinue;
+
+ Vector4f &src = reg(r, temporaryRegister);
+ Int4 condition = As<Int4>(src.x);
+ condition &= r.enableStack[r.enableIndex - 1];
+ r.enableStack[r.enableIndex] = condition;
+
+ Bool notAllFalse = SignMask(condition) != 0;
+ branch(notAllFalse, loopBlock, endBlock);
+
+ Nucleus::setInsertBlock(endBlock);
+ r.enableBreak = restoreBreak;
+
+ Nucleus::setInsertBlock(loopBlock);
+
+ loopRepDepth++;
+ breakDepth = 0;
+ }
+
void PixelRoutine::RET(Registers &r)
{
- if(!returns)
+ if(currentLabel == -1)
{
returnBlock = Nucleus::createBasicBlock();
Nucleus::createBr(returnBlock);
-
- returns = true;
}
else
{
- // FIXME: Encapsulate
- UInt index = r.callStack[--r.stackIndex];
-
llvm::BasicBlock *unreachableBlock = Nucleus::createBasicBlock();
- llvm::Value *value = Nucleus::createLoad(index.address);
- llvm::Value *switchInst = Nucleus::createSwitch(value, unreachableBlock, (int)callRetBlock.size());
- for(unsigned int i = 0; i < callRetBlock.size(); i++)
+ if(callRetBlock[currentLabel].size() > 1) // Pop the return destination from the call stack
{
- Nucleus::addSwitchCase(switchInst, i, callRetBlock[i]);
+ // FIXME: Encapsulate
+ UInt index = r.callStack[--r.stackIndex];
+
+ llvm::Value *value = Nucleus::createLoad(index.address);
+ llvm::Value *switchInst = Nucleus::createSwitch(value, unreachableBlock, (int)callRetBlock[currentLabel].size());
+
+ for(unsigned int i = 0; i < callRetBlock[currentLabel].size(); i++)
+ {
+ Nucleus::addSwitchCase(switchInst, i, callRetBlock[currentLabel][i]);
+ }
+ }
+ else if(callRetBlock[currentLabel].size() == 1) // Jump directly to the unique return destination
+ {
+ Nucleus::createBr(callRetBlock[currentLabel][0]);
+ }
+ else // Function isn't called
+ {
+ Nucleus::createBr(unreachableBlock);
}
Nucleus::setInsertBlock(unreachableBlock);
@@ -5103,30 +5381,38 @@
}
}
- void PixelRoutine::writeDestination(Registers &r, Color4i &d, const Dst &dst)
+ void PixelRoutine::LEAVE(Registers &r)
+ {
+ r.enableLeave = r.enableLeave & ~r.enableStack[r.enableIndex];
+
+ // FIXME: Return from function if all instances left
+ // FIXME: Use enableLeave in other control-flow constructs
+ }
+
+ void PixelRoutine::writeDestination(Registers &r, Vector4i &d, const Dst &dst)
{
switch(dst.type)
{
- case Dst::PARAMETER_TEMP:
+ case Shader::PARAMETER_TEMP:
if(dst.mask & 0x1) r.ri[dst.index].x = d.x;
if(dst.mask & 0x2) r.ri[dst.index].y = d.y;
if(dst.mask & 0x4) r.ri[dst.index].z = d.z;
if(dst.mask & 0x8) r.ri[dst.index].w = d.w;
break;
- case Dst::PARAMETER_INPUT:
+ case Shader::PARAMETER_INPUT:
if(dst.mask & 0x1) r.vi[dst.index].x = d.x;
if(dst.mask & 0x2) r.vi[dst.index].y = d.y;
if(dst.mask & 0x4) r.vi[dst.index].z = d.z;
if(dst.mask & 0x8) r.vi[dst.index].w = d.w;
break;
- case Dst::PARAMETER_CONST: ASSERT(false); break;
- case Dst::PARAMETER_TEXTURE:
+ case Shader::PARAMETER_CONST: ASSERT(false); break;
+ case Shader::PARAMETER_TEXTURE:
if(dst.mask & 0x1) r.ti[dst.index].x = d.x;
if(dst.mask & 0x2) r.ti[dst.index].y = d.y;
if(dst.mask & 0x4) r.ti[dst.index].z = d.z;
if(dst.mask & 0x8) r.ti[dst.index].w = d.w;
break;
- case Dst::PARAMETER_COLOROUT:
+ case Shader::PARAMETER_COLOROUT:
if(dst.mask & 0x1) r.vi[dst.index].x = d.x;
if(dst.mask & 0x2) r.vi[dst.index].y = d.y;
if(dst.mask & 0x4) r.vi[dst.index].z = d.z;
@@ -5137,29 +5423,29 @@
}
}
- Color4i PixelRoutine::regi(Registers &r, const Src &src)
+ Vector4i PixelRoutine::regi(Registers &r, const Src &src)
{
- Color4i *reg;
+ Vector4i *reg;
int i = src.index;
- Color4i c;
+ Vector4i c;
- if(src.type == ShaderParameter::PARAMETER_CONST)
+ if(src.type == Shader::PARAMETER_CONST)
{
- c.r = *Pointer<Short4>(r.data + OFFSET(DrawData,ps.cW[i][0]));
- c.g = *Pointer<Short4>(r.data + OFFSET(DrawData,ps.cW[i][1]));
- c.b = *Pointer<Short4>(r.data + OFFSET(DrawData,ps.cW[i][2]));
- c.a = *Pointer<Short4>(r.data + OFFSET(DrawData,ps.cW[i][3]));
+ 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]));
}
switch(src.type)
{
- case Src::PARAMETER_TEMP: reg = &r.ri[i]; break;
- case Src::PARAMETER_INPUT: reg = &r.vi[i]; break;
- case Src::PARAMETER_CONST: reg = &c; break;
- case Src::PARAMETER_TEXTURE: reg = &r.ti[i]; break;
- case Src::PARAMETER_VOID: return r.ri[0]; // Dummy
- case Src::PARAMETER_FLOATLITERAL: return r.ri[0]; // Dummy
+ case Shader::PARAMETER_TEMP: reg = &r.ri[i]; break;
+ case Shader::PARAMETER_INPUT: reg = &r.vi[i]; break;
+ case Shader::PARAMETER_CONST: reg = &c; break;
+ case Shader::PARAMETER_TEXTURE: reg = &r.ti[i]; break;
+ case Shader::PARAMETER_VOID: return r.ri[0]; // Dummy
+ case Shader::PARAMETER_FLOAT4LITERAL: return r.ri[0]; // Dummy
default:
ASSERT(false);
}
@@ -5169,180 +5455,177 @@
Short4 &z = (*reg)[(src.swizzle >> 4) & 0x3];
Short4 &w = (*reg)[(src.swizzle >> 6) & 0x3];
- Color4i mod;
+ Vector4i mod;
switch(src.modifier)
{
- case Src::MODIFIER_NONE:
- mod.r = x;
- mod.g = y;
- mod.b = z;
- mod.a = w;
+ case Shader::MODIFIER_NONE:
+ mod.x = x;
+ mod.y = y;
+ mod.z = z;
+ mod.w = w;
break;
- case Src::MODIFIER_BIAS:
- mod.r = SubSat(x, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- mod.g = SubSat(y, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- mod.b = SubSat(z, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- mod.a = SubSat(w, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ case Shader::MODIFIER_BIAS:
+ mod.x = SubSat(x, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ mod.y = SubSat(y, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ mod.z = SubSat(z, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ mod.w = SubSat(w, Short4(0x0800, 0x0800, 0x0800, 0x0800));
break;
- case Src::MODIFIER_BIAS_NEGATE:
- mod.r = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), x);
- mod.g = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), y);
- mod.b = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), z);
- mod.a = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), w);
+ case Shader::MODIFIER_BIAS_NEGATE:
+ mod.x = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), x);
+ mod.y = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), y);
+ mod.z = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), z);
+ mod.w = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), w);
break;
- case Src::MODIFIER_COMPLEMENT:
- mod.r = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), x);
- mod.g = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), y);
- mod.b = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), z);
- mod.a = SubSat(Short4(0x1000, 0x1000, 0x1000, 0x1000), w);
+ case Shader::MODIFIER_COMPLEMENT:
+ mod.x = SubSat(Short4(0x1000), x);
+ mod.y = SubSat(Short4(0x1000), y);
+ mod.z = SubSat(Short4(0x1000), z);
+ mod.w = SubSat(Short4(0x1000), w);
break;
- case Src::MODIFIER_NEGATE:
- mod.r = -x;
- mod.g = -y;
- mod.b = -z;
- mod.a = -w;
+ case Shader::MODIFIER_NEGATE:
+ mod.x = -x;
+ mod.y = -y;
+ mod.z = -z;
+ mod.w = -w;
break;
- case Src::MODIFIER_X2:
- mod.r = AddSat(x, x);
- mod.g = AddSat(y, y);
- mod.b = AddSat(z, z);
- mod.a = AddSat(w, w);
+ case Shader::MODIFIER_X2:
+ mod.x = AddSat(x, x);
+ mod.y = AddSat(y, y);
+ mod.z = AddSat(z, z);
+ mod.w = AddSat(w, w);
break;
- case Src::MODIFIER_X2_NEGATE:
- mod.r = -AddSat(x, x);
- mod.g = -AddSat(y, y);
- mod.b = -AddSat(z, z);
- mod.a = -AddSat(w, w);
+ case Shader::MODIFIER_X2_NEGATE:
+ mod.x = -AddSat(x, x);
+ mod.y = -AddSat(y, y);
+ mod.z = -AddSat(z, z);
+ mod.w = -AddSat(w, w);
break;
- case Src::MODIFIER_SIGN:
- mod.r = SubSat(x, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- mod.g = SubSat(y, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- mod.b = SubSat(z, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- mod.a = SubSat(w, Short4(0x0800, 0x0800, 0x0800, 0x0800));
- mod.r = AddSat(mod.r, mod.r);
- mod.g = AddSat(mod.g, mod.g);
- mod.b = AddSat(mod.b, mod.b);
- mod.a = AddSat(mod.a, mod.a);
+ case Shader::MODIFIER_SIGN:
+ mod.x = SubSat(x, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ mod.y = SubSat(y, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ mod.z = SubSat(z, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ mod.w = SubSat(w, Short4(0x0800, 0x0800, 0x0800, 0x0800));
+ mod.x = AddSat(mod.x, mod.x);
+ mod.y = AddSat(mod.y, mod.y);
+ mod.z = AddSat(mod.z, mod.z);
+ mod.w = AddSat(mod.w, mod.w);
break;
- case Src::MODIFIER_SIGN_NEGATE:
- mod.r = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), x);
- mod.g = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), y);
- mod.b = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), z);
- mod.a = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), w);
- mod.r = AddSat(mod.r, mod.r);
- mod.g = AddSat(mod.g, mod.g);
- mod.b = AddSat(mod.b, mod.b);
- mod.a = AddSat(mod.a, mod.a);
+ case Shader::MODIFIER_SIGN_NEGATE:
+ mod.x = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), x);
+ mod.y = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), y);
+ mod.z = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), z);
+ mod.w = SubSat(Short4(0x0800, 0x0800, 0x0800, 0x0800), w);
+ mod.x = AddSat(mod.x, mod.x);
+ mod.y = AddSat(mod.y, mod.y);
+ mod.z = AddSat(mod.z, mod.z);
+ mod.w = AddSat(mod.w, mod.w);
break;
- case Src::MODIFIER_DZ:
- mod.r = x;
- mod.g = y;
- mod.b = z;
- mod.a = w;
+ case Shader::MODIFIER_DZ:
+ mod.x = x;
+ mod.y = y;
+ mod.z = z;
+ mod.w = w;
// Projection performed by texture sampler
break;
- case Src::MODIFIER_DW:
- mod.r = x;
- mod.g = y;
- mod.b = z;
- mod.a = w;
+ case Shader::MODIFIER_DW:
+ mod.x = x;
+ mod.y = y;
+ mod.z = z;
+ mod.w = w;
// Projection performed by texture sampler
break;
default:
ASSERT(false);
}
- if(src.type == ShaderParameter::PARAMETER_CONST && (src.modifier == Src::MODIFIER_X2 || src.modifier == Src::MODIFIER_X2_NEGATE))
+ if(src.type == Shader::PARAMETER_CONST && (src.modifier == Shader::MODIFIER_X2 || src.modifier == Shader::MODIFIER_X2_NEGATE))
{
- mod.r = Min(mod.r, Short4(0x1000, 0x1000, 0x1000, 0x1000)); mod.r = Max(mod.r, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
- mod.g = Min(mod.g, Short4(0x1000, 0x1000, 0x1000, 0x1000)); mod.g = Max(mod.g, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
- mod.b = Min(mod.b, Short4(0x1000, 0x1000, 0x1000, 0x1000)); mod.b = Max(mod.b, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
- mod.a = Min(mod.a, Short4(0x1000, 0x1000, 0x1000, 0x1000)); mod.a = Max(mod.a, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
+ mod.x = Min(mod.x, Short4(0x1000)); mod.x = Max(mod.x, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
+ mod.y = Min(mod.y, Short4(0x1000)); mod.y = Max(mod.y, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
+ mod.z = Min(mod.z, Short4(0x1000)); mod.z = Max(mod.z, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
+ mod.w = Min(mod.w, Short4(0x1000)); mod.w = Max(mod.w, Short4(-0x1000, -0x1000, -0x1000, -0x1000));
}
return mod;
}
- Color4f PixelRoutine::reg(Registers &r, const Src &src, int offset)
+ Vector4f PixelRoutine::reg(Registers &r, const Src &src, int offset)
{
- Color4f reg;
+ Vector4f reg;
int i = src.index + offset;
switch(src.type)
{
- case Src::PARAMETER_TEMP: reg = r.rf[i]; break;
- case Src::PARAMETER_INPUT:
+ case Shader::PARAMETER_TEMP:
+ if(src.rel.type == Shader::PARAMETER_VOID)
{
- if(!src.relative)
+ reg = r.rf[i];
+ }
+ else
+ {
+ Int a = relativeAddress(r, src);
+
+ reg = r.rf[i + a];
+ }
+ break;
+ case Shader::PARAMETER_INPUT:
+ {
+ if(src.rel.type == Shader::PARAMETER_VOID) // Not relative
{
- reg.x = r.vx[i];
- reg.y = r.vy[i];
- reg.z = r.vz[i];
- reg.w = r.vw[i];
+ reg = r.vf[i];
}
- else if(src.relativeType == Src::PARAMETER_LOOP)
+ else if(src.rel.type == Shader::PARAMETER_LOOP)
{
Int aL = r.aL[r.loopDepth];
- reg.x = r.vx[i + aL];
- reg.y = r.vy[i + aL];
- reg.z = r.vz[i + aL];
- reg.w = r.vw[i + aL];
+ reg = r.vf[i + aL];
}
- else ASSERT(false);
- }
- break;
- case Src::PARAMETER_CONST:
- {
- reg.r = reg.g = reg.b = reg.a = *Pointer<Float4>(r.data + OFFSET(DrawData,ps.c[i]));
-
- reg.r = reg.r.xxxx;
- reg.g = reg.g.yyyy;
- reg.b = reg.b.zzzz;
- reg.a = reg.a.wwww;
-
- if(localShaderConstants) // Constant may be known at compile time
+ else
{
- for(int j = 0; j < pixelShader->getLength(); j++)
- {
- const ShaderInstruction &instruction = *pixelShader->getInstruction(j);
-
- if(instruction.getOpcode() == ShaderOperation::OPCODE_DEF)
- {
- if(instruction.getDestinationParameter().index == i)
- {
- reg.r = Float4(instruction.getSourceParameter(0).value);
- reg.g = Float4(instruction.getSourceParameter(1).value);
- reg.b = Float4(instruction.getSourceParameter(2).value);
- reg.a = Float4(instruction.getSourceParameter(3).value);
-
- break;
- }
- }
- }
+ Int a = relativeAddress(r, src);
+
+ reg = r.vf[i + a];
}
}
break;
- case Src::PARAMETER_TEXTURE:
- {
- reg.x = r.vx[2 + i];
- reg.y = r.vy[2 + i];
- reg.z = r.vz[2 + i];
- reg.w = r.vw[2 + i];
- }
+ case Shader::PARAMETER_CONST:
+ reg = readConstant(r, src, offset);
break;
- case Src::PARAMETER_MISCTYPE:
+ case Shader::PARAMETER_TEXTURE:
+ reg = r.vf[2 + i];
+ break;
+ case Shader::PARAMETER_MISCTYPE:
if(src.index == 0) reg = r.vPos;
if(src.index == 1) reg = r.vFace;
break;
- case Src::PARAMETER_SAMPLER: return r.rf[0]; // Dummy
- case Src::PARAMETER_PREDICATE: return r.rf[0]; // Dummy
- case Src::PARAMETER_VOID: return r.rf[0]; // Dummy
- case Src::PARAMETER_FLOATLITERAL: return r.rf[0]; // Dummy
- case Src::PARAMETER_CONSTINT: return r.rf[0]; // Dummy
- case Src::PARAMETER_CONSTBOOL: return r.rf[0]; // Dummy
- case Src::PARAMETER_LOOP: return r.rf[0]; // Dummy
+ case Shader::PARAMETER_SAMPLER:
+ if(src.rel.type == Shader::PARAMETER_VOID)
+ {
+ reg.x = As<Float4>(Int4(i));
+ }
+ else if(src.rel.type == Shader::PARAMETER_TEMP)
+ {
+ reg.x = As<Float4>(Int4(i) + RoundInt(r.rf[src.rel.index].x));
+ }
+ return reg;
+ case Shader::PARAMETER_PREDICATE: return reg; // Dummy
+ case Shader::PARAMETER_VOID: return reg; // Dummy
+ case Shader::PARAMETER_FLOAT4LITERAL:
+ reg.x = Float4(src.value[0]);
+ reg.y = Float4(src.value[1]);
+ reg.z = Float4(src.value[2]);
+ reg.w = Float4(src.value[3]);
+ break;
+ case Shader::PARAMETER_CONSTINT: return reg; // Dummy
+ case Shader::PARAMETER_CONSTBOOL: return reg; // Dummy
+ case Shader::PARAMETER_LOOP: return reg; // Dummy
+ case Shader::PARAMETER_COLOROUT:
+ reg = r.oC[i];
+ break;
+ case Shader::PARAMETER_DEPTHOUT:
+ reg.x = r.oDepth;
+ break;
default:
ASSERT(false);
}
@@ -5352,29 +5635,29 @@
Float4 &z = reg[(src.swizzle >> 4) & 0x3];
Float4 &w = reg[(src.swizzle >> 6) & 0x3];
- Color4f mod;
+ Vector4f mod;
switch(src.modifier)
{
- case Src::MODIFIER_NONE:
+ case Shader::MODIFIER_NONE:
mod.x = x;
mod.y = y;
mod.z = z;
mod.w = w;
break;
- case Src::MODIFIER_NEGATE:
+ case Shader::MODIFIER_NEGATE:
mod.x = -x;
mod.y = -y;
mod.z = -z;
mod.w = -w;
break;
- case Src::MODIFIER_ABS:
+ case Shader::MODIFIER_ABS:
mod.x = Abs(x);
mod.y = Abs(y);
mod.z = Abs(z);
mod.w = Abs(w);
break;
- case Src::MODIFIER_ABS_NEGATE:
+ case Shader::MODIFIER_ABS_NEGATE:
mod.x = -Abs(x);
mod.y = -Abs(y);
mod.z = -Abs(z);
@@ -5387,13 +5670,134 @@
return mod;
}
- bool PixelRoutine::colorUsed()
+ Vector4f PixelRoutine::readConstant(Registers &r, const Src &src, int offset)
{
- return state.colorWriteMask || state.alphaTestActive() || state.shaderContainsTexkill;
+ Vector4f c;
+
+ 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.x.xxxx;
+ c.y = c.y.yyyy;
+ c.z = c.z.zzzz;
+ c.w = c.w.wwww;
+
+ if(localShaderConstants) // Constant may be known at compile time
+ {
+ for(int j = 0; j < shader->getLength(); j++)
+ {
+ const Shader::Instruction &instruction = *shader->getInstruction(j);
+
+ if(instruction.opcode == Shader::OPCODE_DEF)
+ {
+ if(instruction.dst.index == i)
+ {
+ c.x = Float4(instruction.src[0].value[0]);
+ c.y = Float4(instruction.src[0].value[1]);
+ c.z = Float4(instruction.src[0].value[2]);
+ c.w = Float4(instruction.src[0].value[3]);
+
+ break;
+ }
+ }
+ }
+ }
+ }
+ else if(src.rel.type == Shader::PARAMETER_LOOP)
+ {
+ Int loopCounter = r.aL[r.loopDepth];
+
+ c.x = c.y = c.z = c.w = *Pointer<Float4>(r.data + OFFSET(DrawData,ps.c[i]) + loopCounter * 16);
+
+ c.x = c.x.xxxx;
+ c.y = c.y.yyyy;
+ c.z = c.z.zzzz;
+ c.w = c.w.wwww;
+ }
+ else
+ {
+ Int a = relativeAddress(r, src);
+
+ c.x = c.y = c.z = c.w = *Pointer<Float4>(r.data + OFFSET(DrawData,ps.c[i]) + a * 16);
+
+ c.x = c.x.xxxx;
+ c.y = c.y.yyyy;
+ c.z = c.z.zzzz;
+ c.w = c.w.wwww;
+ }
+
+ return c;
}
- unsigned short PixelRoutine::pixelShaderVersion() const
+ Int PixelRoutine::relativeAddress(Registers &r, const Shader::Parameter &var)
{
- return pixelShader ? pixelShader->getVersion() : 0x0000;
+ ASSERT(var.rel.deterministic);
+
+ if(var.rel.type == Shader::PARAMETER_TEMP)
+ {
+ return RoundInt(Extract(r.rf[var.rel.index].x, 0)) * var.rel.scale;
+ }
+ else if(var.rel.type == Shader::PARAMETER_INPUT)
+ {
+ return RoundInt(Extract(r.vf[var.rel.index].x, 0)) * var.rel.scale;
+ }
+ else if(var.rel.type == Shader::PARAMETER_OUTPUT)
+ {
+ return RoundInt(Extract(r.oC[var.rel.index].x, 0)) * var.rel.scale;
+ }
+ else if(var.rel.type == Shader::PARAMETER_CONST)
+ {
+ RValue<Float4> c = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c[var.rel.index]));
+
+ return RoundInt(Extract(c, 0)) * var.rel.scale;
+ }
+ else ASSERT(false);
+
+ return 0;
+ }
+
+ Int4 PixelRoutine::enableMask(Registers &r, const Shader::Instruction *instruction)
+ {
+ Int4 enable = instruction->analysisBranch ? Int4(r.enableStack[r.enableIndex]) : Int4(0xFFFFFFFF);
+
+ if(shader->containsBreakInstruction() && !whileTest && instruction->analysisBreak)
+ {
+ enable &= r.enableBreak;
+ }
+
+ if(shader->containsContinueInstruction() && !whileTest && instruction->analysisContinue)
+ {
+ enable &= r.enableContinue;
+ }
+
+ if(shader->containsLeaveInstruction() && instruction->analysisLeave)
+ {
+ enable &= r.enableLeave;
+ }
+
+ return enable;
+ }
+
+ bool PixelRoutine::colorUsed()
+ {
+ return state.colorWriteMask || state.alphaTestActive() || state.shaderContainsKill;
+ }
+
+ unsigned short PixelRoutine::shaderVersion() const
+ {
+ return shader ? shader->getVersion() : 0x0000;
+ }
+
+ bool PixelRoutine::interpolateZ() const
+ {
+ return state.depthTestActive || state.pixelFogActive() || (shader && shader->vPosDeclared && fullPixelPositionRegister);
+ }
+
+ bool PixelRoutine::interpolateW() const
+ {
+ return state.perspective || (shader && shader->vPosDeclared && fullPixelPositionRegister);
}
}
diff --git a/src/Shader/PixelRoutine.hpp b/src/Shader/PixelRoutine.hpp
index c1070ff..29ea75a 100644
--- a/src/Shader/PixelRoutine.hpp
+++ b/src/Shader/PixelRoutine.hpp
@@ -1,304 +1,327 @@
-// SwiftShader Software Renderer
-//
-// Copyright(c) 2005-2011 TransGaming Inc.
-//
-// All rights reserved. No part of this software may be copied, distributed, transmitted,
-// transcribed, stored in a retrieval system, translated into any human or computer
-// language by any means, or disclosed to third parties without the explicit written
-// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
-// or implied, including but not limited to any patent rights, are granted to you.
-//
-
-#ifndef sw_PixelRoutine_hpp
-#define sw_PixelRoutine_hpp
-
-#include "Rasterizer.hpp"
-#include "ShaderCore.hpp"
-
-#include "Types.hpp"
-
-namespace sw
-{
- extern bool forceClearRegisters;
-
- class PixelShader;
- class SamplerCore;
-
- class PixelRoutine : public Rasterizer, public ShaderCore
- {
- friend PixelProcessor; // FIXME
-
- public:
- PixelRoutine(const PixelProcessor::State &state, const PixelShader *pixelShader);
-
- ~PixelRoutine();
-
- protected:
- struct Registers
- {
- Registers() : current(ri[0]), diffuse(vi[0]), specular(vi[1]), callStack(4), aL(4), increment(4), iteration(4), enableStack(1 + 24), vx(10), vy(10), vz(10), vw(10)
- {
- if(forceClearRegisters)
- {
- for(int i = 0; i < 10; i++)
- {
- vx[i] = Float4(0, 0, 0, 0);
- vy[i] = Float4(0, 0, 0, 0);
- vz[i] = Float4(0, 0, 0, 0);
- vw[i] = Float4(0, 0, 0, 0);
- }
-
- for(int i = 0; i < 4; i++)
- {
- oC[i].r = Float4(0.0f);
- oC[i].g = Float4(0.0f);
- oC[i].b = Float4(0.0f);
- oC[i].a = Float4(0.0f);
- }
- }
-
- loopDepth = -1;
- enableStack[0] = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
- enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
-
- occlusion = 0;
-
- #if PERF_PROFILE
- for(int i = 0; i < PERF_TIMERS; i++)
- {
- cycles[i] = 0;
- }
- #endif
- }
-
- Pointer<Byte> constants;
-
- Pointer<Byte> primitive;
- Int cluster;
- Pointer<Byte> data;
-
- Float4 z[4];
- Float4 rhw;
-
- Float4 Dz[4];
- Float4 Dw;
- Float4 Dv[10][4];
- Float4 Df;
-
- Color4i ¤t;
- Color4i &diffuse;
- Color4i &specular;
-
- Color4i ri[6];
- Color4i vi[2];
- Color4i ti[6];
-
- Color4f rf[32];
- Array<Float4> vx;
- Array<Float4> vy;
- Array<Float4> vz;
- Array<Float4> vw;
-
- Color4f vPos;
- Color4f vFace;
-
- Color4f oC[4];
- Float4 oDepth;
-
- Color4f p0;
- Array<Int> aL;
-
- Array<Int> increment;
- Array<Int> iteration;
-
- Int loopDepth;
- Int stackIndex; // FIXME: Inc/decrement callStack
- Array<UInt> callStack;
-
- Int enableIndex;
- Array<Int4> enableStack;
- Int4 enableBreak;
-
- // 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
-
- UInt occlusion;
-
- #if PERF_PROFILE
- Long cycles[PERF_TIMERS];
- #endif
- };
-
- void quad(Registers &r, Pointer<Byte> cBuffer[4], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y);
-
- Float4 interpolate(Float4 &x, Float4 &D, Float4 &rhw, Pointer<Byte> planeEquation, bool flat, bool perspective);
- 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, Context::StencilCompareMode stencilCompareMode, bool CCW);
- void stencilOperation(Registers &r, Byte8 &newValue, Byte8 &bufferValue, Context::StencilOperation stencilPassOperation, Context::StencilOperation stencilZFailOperation, Context::StencilOperation stencilFailOperation, bool CCW, Int &zMask, Int &sMask);
- void stencilOperation(Registers &r, Byte8 &output, Byte8 &bufferValue, Context::StencilOperation operation, bool CCW);
- Bool depthTest(Registers &r, Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &sMask, Int &zMask, Int &cMask);
- void blendTexture(Registers &r, Color4i ¤t, Color4i &temp, Color4i &texture, int stage);
- void alphaTest(Registers &r, Int &aMask, Short4 &alpha);
- void alphaToCoverage(Registers &r, Int cMask[4], Float4 &alpha);
- Bool alphaTest(Registers &r, Int cMask[4], Color4i ¤t);
- Bool alphaTest(Registers &r, Int cMask[4], Color4f &c0);
- void fogBlend(Registers &r, Color4i ¤t, Float4 &fog, Float4 &z, Float4 &rhw);
- void fogBlend(Registers &r, Color4f &c0, Float4 &fog, Float4 &z, Float4 &rhw);
- void pixelFog(Registers &r, Float4 &visibility, Float4 &z, Float4 &rhw);
- void specularPixel(Color4i ¤t, Color4i &specular);
-
- void sampleTexture(Registers &r, Color4i &c, int coordinates, int sampler, bool project = false);
- void sampleTexture(Registers &r, Color4i &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project = false, bool bias = false, bool fixed12 = true);
- void sampleTexture(Registers &r, Color4i &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool project = false, bool bias = false, bool fixed12 = true, bool gradients = false, bool lodProvided = false);
- void sampleTexture(Registers &r, Color4f &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool project = false, bool bias = false, bool gradients = false, bool lodProvided = false);
-
- // Raster operations
- void clampColor(Color4f oC[4]);
- void rasterOperation(Color4i ¤t, Registers &r, Float4 &fog, Pointer<Byte> &cBuffer, Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
- void rasterOperation(Color4f oC[4], Registers &r, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
- void blendFactor(Registers &r, const Color4i &blendFactor, const Color4i ¤t, const Color4i &pixel, Context::BlendFactor blendFactorActive);
- void blendFactorAlpha(Registers &r, const Color4i &blendFactor, const Color4i ¤t, const Color4i &pixel, Context::BlendFactor blendFactorAlphaActive);
- void alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Color4i ¤t, Int &x);
- void writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &i, Color4i ¤t, Int &sMask, Int &zMask, Int &cMask);
- void blendFactor(Registers &r, const Color4f &blendFactor, const Color4f &oC, const Color4f &pixel, Context::BlendFactor blendFactorActive);
- void blendFactorAlpha(Registers &r, const Color4f &blendFactor, const Color4f &oC, const Color4f &pixel, Context::BlendFactor blendFactorAlphaActive);
- void alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Color4f &oC, Int &x);
- void writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &i, Color4f &oC, Int &sMask, Int &zMask, Int &cMask);
- 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 ps_1_x(Registers &r, Int cMask[4]);
- void ps_2_x(Registers &r, Int cMask[4]);
-
- Short4 convertFixed12(Float4 &cf);
- void convertFixed12(Color4i &ci, Color4f &cf);
- Float4 convertSigned12(Short4 &ci);
- void convertSigned12(Color4f &cf, Color4i &ci);
- Float4 convertUnsigned16(UShort4 ci);
- UShort4 convertFixed16(Float4 &cf, bool saturate = true);
- void convertFixed16(Color4i &ci, Color4f &cf, bool saturate = true);
- void sRGBtoLinear16_16(Registers &r, Color4i &c);
- void sRGBtoLinear12_16(Registers &r, Color4i &c);
- void linearToSRGB16_16(Registers &r, Color4i &c);
- void linearToSRGB12_16(Registers &r, Color4i &c);
- Float4 sRGBtoLinear(const Float4 &x);
- Float4 linearToSRGB(const Float4 &x);
-
- typedef Shader::Instruction::DestinationParameter Dst;
- typedef Shader::Instruction::SourceParameter Src;
- typedef Shader::Instruction::Operation Op;
- typedef Shader::Instruction::Operation::Control Control;
-
- // ps_1_x instructions
- void MOV(Color4i &dst, Color4i &src0);
- void ADD(Color4i &dst, Color4i &src0, Color4i &src1);
- void SUB(Color4i &dst, Color4i &src0, Color4i &src1);
- void MAD(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2);
- void MUL(Color4i &dst, Color4i &src0, Color4i &src1);
- void DP3(Color4i &dst, Color4i &src0, Color4i &src1);
- void DP4(Color4i &dst, Color4i &src0, Color4i &src1);
- void LRP(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2);
- void TEXCOORD(Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate);
- void TEXCRD(Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate, bool project);
- void TEXDP3(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, Color4i &src);
- void TEXDP3TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0);
- void TEXKILL(Int cMask[4], Float4 &u, Float4 &v, Float4 &s);
- void TEXKILL(Int cMask[4], Color4i &dst);
- void TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, bool project);
- void TEXLD(Registers &r, Color4i &dst, Color4i &src, int stage, bool project);
- void TEXBEM(Registers &r, Color4i &dst, Color4i &src, Float4 &u, Float4 &v, Float4 &s, int stage);
- void TEXBEML(Registers &r, Color4i &dst, Color4i &src, Float4 &u, Float4 &v, Float4 &s, int stage);
- void TEXREG2AR(Registers &r, Color4i &dst, Color4i &src0, int stage);
- void TEXREG2GB(Registers &r, Color4i &dst, Color4i &src0, int stage);
- void TEXREG2RGB(Registers &r, Color4i &dst, Color4i &src0, int stage);
- void TEXM3X2DEPTH(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, Color4i &src, bool signedScaling);
- void TEXM3X2PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Color4i &src0, int component, bool signedScaling);
- void TEXM3X2TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0, bool signedScaling);
- void TEXM3X3(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, Color4i &src0, bool signedScaling);
- void TEXM3X3PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Color4i &src0, int component, bool signedScaling);
- void TEXM3X3SPEC(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0, Color4i &src1);
- void TEXM3X3TEX(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0, bool singedScaling);
- void TEXM3X3VSPEC(Registers &r, Color4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Color4i &src0);
- void TEXDEPTH(Registers &r);
- void CND(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2);
- void CMP(Color4i &dst, Color4i &src0, Color4i &src1, Color4i &src2);
- void BEM(Registers &r, Color4i &dst, Color4i &src0, Color4i &src1, int stage);
-
- // ps_2_x instructions
- void M3X2(Registers &r, Color4f &dst, Color4f &src0, const Src &src1);
- void M3X3(Registers &r, Color4f &dst, Color4f &src0, const Src &src1);
- void M3X4(Registers &r, Color4f &dst, Color4f &src0, const Src &src1);
- void M4X3(Registers &r, Color4f &dst, Color4f &src0, const Src &src1);
- void M4X4(Registers &r, Color4f &dst, Color4f &src0, const Src &src1);
- void TEXLD(Registers &r, Color4f &dst, Color4f &src0, const Src &src1, bool project, bool bias);
- void TEXLDD(Registers &r, Color4f &dst, Color4f &src0, const Src &src1, Color4f &src2, Color4f &src3, bool project, bool bias);
- void TEXLDL(Registers &r, Color4f &dst, Color4f &src0, const Src &src1, bool project, bool bias);
- void TEXKILL(Int cMask[4], Color4f &src, unsigned char mask);
- void DSX(Color4f &dst, Color4f &src);
- void DSY(Color4f &dst, Color4f &src);
- void BREAK(Registers &r);
- void BREAKC(Registers &r, Color4f &src0, Color4f &src1, Control);
- void BREAKP(Registers &r, const Src &predicateRegister);
- void CALL(Registers &r, int labelIndex);
- void CALLNZ(Registers &r, int labelIndex, const Src &src);
- void CALLNZb(Registers &r, int labelIndex, const Src &boolRegister);
- void CALLNZp(Registers &r, int labelIndex, const Src &predicateRegister);
- void ELSE(Registers &r);
- void ENDIF(Registers &r);
- void ENDLOOP(Registers &r);
- void ENDREP(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, Color4f &src0, Color4f &src1, Control);
- void LABEL(int labelIndex);
- void LOOP(Registers &r, const Src &integerRegister);
- void REP(Registers &r, const Src &integerRegister);
- void RET(Registers &r);
-
- void readConstant(Registers &r, int index);
-
- void writeDestination(Registers &r, Color4i &d, const Dst &dst);
- Color4i regi(Registers &r, const Src &src);
- Color4f reg(Registers &r, const Src &src, int offset = 0);
-
- bool colorUsed();
- unsigned short pixelShaderVersion() const;
-
- private:
- SamplerCore *sampler[16];
-
- bool perturbate;
- bool luminance;
- bool previousScaling;
-
- bool returns;
- int ifDepth;
- int loopRepDepth;
- int breakDepth;
-
- // 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;
- llvm::BasicBlock *returnBlock;
- bool isConditionalIf[24 + 24];
-
- const PixelShader *const pixelShader;
- };
-}
-
-#endif // sw_PixelRoutine_hpp
+// SwiftShader Software Renderer
+//
+// Copyright(c) 2005-2012 TransGaming Inc.
+//
+// All rights reserved. No part of this software may be copied, distributed, transmitted,
+// transcribed, stored in a retrieval system, translated into any human or computer
+// language by any means, or disclosed to third parties without the explicit written
+// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
+// or implied, including but not limited to any patent rights, are granted to you.
+//
+
+#ifndef sw_PixelRoutine_hpp
+#define sw_PixelRoutine_hpp
+
+#include "Rasterizer.hpp"
+#include "ShaderCore.hpp"
+#include "PixelShader.hpp"
+
+#include "Types.hpp"
+
+namespace sw
+{
+ extern bool forceClearRegisters;
+
+ class PixelShader;
+ class SamplerCore;
+
+ class PixelRoutine : public Rasterizer, public ShaderCore
+ {
+ friend PixelProcessor; // FIXME
+
+ public:
+ PixelRoutine(const PixelProcessor::State &state, const PixelShader *shader);
+
+ ~PixelRoutine();
+
+ protected:
+ struct Registers
+ {
+ Registers(const PixelShader *shader) :
+ current(ri[0]), diffuse(vi[0]), specular(vi[1]),
+ rf(shader && shader->dynamicallyIndexedTemporaries),
+ vf(shader && shader->dynamicallyIndexedInput)
+ {
+ if(!shader || shader->getVersion() < 0x0200 || forceClearRegisters)
+ {
+ for(int i = 0; i < 10; i++)
+ {
+ vf[i].x = Float4(0.0f);
+ vf[i].y = Float4(0.0f);
+ vf[i].z = Float4(0.0f);
+ vf[i].w = Float4(0.0f);
+ }
+ }
+
+ 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);
+ }
+
+ if(shader && shader->containsLeaveInstruction())
+ {
+ enableLeave = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+ }
+
+ occlusion = 0;
+
+ #if PERF_PROFILE
+ for(int i = 0; i < PERF_TIMERS; i++)
+ {
+ cycles[i] = 0;
+ }
+ #endif
+ }
+
+ Pointer<Byte> constants;
+
+ Pointer<Byte> primitive;
+ Int cluster;
+ Pointer<Byte> data;
+
+ Float4 z[4];
+ Float4 w;
+ Float4 rhw;
+
+ Float4 Dz[4];
+ Float4 Dw;
+ Float4 Dv[10][4];
+ Float4 Df;
+
+ Vector4i ¤t;
+ Vector4i &diffuse;
+ Vector4i &specular;
+
+ Vector4i ri[6];
+ Vector4i vi[2];
+ Vector4i ti[6];
+
+ RegisterArray<4096> rf;
+ RegisterArray<10> vf;
+
+ Vector4f vPos;
+ Vector4f vFace;
+
+ Vector4f oC[4];
+ Float4 oDepth;
+
+ Vector4f p0;
+ Array<Int, 4> aL;
+
+ Array<Int, 4> increment;
+ Array<Int, 4> iteration;
+
+ Int loopDepth;
+ Int stackIndex; // FIXME: Inc/decrement callStack
+ Array<UInt, 4> callStack;
+
+ Int enableIndex;
+ Array<Int4, 1 + 24> enableStack;
+ Int4 enableBreak;
+ Int4 enableContinue;
+ Int4 enableLeave;
+
+ // 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
+
+ UInt occlusion;
+
+ #if PERF_PROFILE
+ Long cycles[PERF_TIMERS];
+ #endif
+ };
+
+ typedef Shader::DestinationParameter Dst;
+ typedef Shader::SourceParameter Src;
+ typedef Shader::Control Control;
+
+ void quad(Registers &r, Pointer<Byte> cBuffer[4], Pointer<Byte> &zBuffer, Pointer<Byte> &sBuffer, Int cMask[4], Int &x, Int &y);
+
+ Float4 interpolate(Float4 &x, Float4 &D, Float4 &rhw, Pointer<Byte> planeEquation, bool flat, bool perspective);
+ 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, Context::StencilCompareMode stencilCompareMode, bool CCW);
+ void stencilOperation(Registers &r, Byte8 &newValue, Byte8 &bufferValue, Context::StencilOperation stencilPassOperation, Context::StencilOperation stencilZFailOperation, Context::StencilOperation stencilFailOperation, bool CCW, Int &zMask, Int &sMask);
+ void stencilOperation(Registers &r, Byte8 &output, Byte8 &bufferValue, Context::StencilOperation operation, bool CCW);
+ Bool depthTest(Registers &r, Pointer<Byte> &zBuffer, int q, Int &x, Float4 &z, Int &sMask, Int &zMask, Int &cMask);
+ void blendTexture(Registers &r, Vector4i ¤t, Vector4i &temp, Vector4i &texture, int stage);
+ void alphaTest(Registers &r, Int &aMask, Short4 &alpha);
+ void alphaToCoverage(Registers &r, Int cMask[4], Float4 &alpha);
+ Bool alphaTest(Registers &r, Int cMask[4], Vector4i ¤t);
+ Bool alphaTest(Registers &r, Int cMask[4], Vector4f &c0);
+ void fogBlend(Registers &r, Vector4i ¤t, Float4 &fog, Float4 &z, Float4 &rhw);
+ void fogBlend(Registers &r, Vector4f &c0, Float4 &fog, Float4 &z, Float4 &rhw);
+ void pixelFog(Registers &r, Float4 &visibility, Float4 &z, Float4 &rhw);
+ void specularPixel(Vector4i ¤t, Vector4i &specular);
+
+ void sampleTexture(Registers &r, Vector4i &c, int coordinates, int sampler, bool project = false);
+ void sampleTexture(Registers &r, Vector4i &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project = false, bool bias = false, bool fixed12 = true);
+ void sampleTexture(Registers &r, Vector4i &c, int sampler, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool project = false, bool bias = false, bool fixed12 = true, bool gradients = false, bool lodProvided = false);
+ 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[4]);
+ void rasterOperation(Vector4i ¤t, Registers &r, Float4 &fog, Pointer<Byte> &cBuffer, Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
+ void rasterOperation(Vector4f oC[4], Registers &r, Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]);
+ void blendFactor(Registers &r, const Vector4i &blendFactor, const Vector4i ¤t, const Vector4i &pixel, Context::BlendFactor blendFactorActive);
+ void blendFactorAlpha(Registers &r, const Vector4i &blendFactor, const Vector4i ¤t, const Vector4i &pixel, Context::BlendFactor blendFactorAlphaActive);
+ void alphaBlend(Registers &r, int index, Pointer<Byte> &cBuffer, Vector4i ¤t, Int &x);
+ void writeColor(Registers &r, int index, Pointer<Byte> &cBuffer, Int &i, Vector4i ¤t, Int &sMask, Int &zMask, Int &cMask);
+ void blendFactor(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, Context::BlendFactor blendFactorActive);
+ void blendFactorAlpha(Registers &r, const Vector4f &blendFactor, const Vector4f &oC, const Vector4f &pixel, Context::BlendFactor blendFactorAlphaActive);
+ 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 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 ps_1_x(Registers &r, Int cMask[4]);
+ void ps_2_x(Registers &r, Int cMask[4]);
+
+ Short4 convertFixed12(RValue<Float4> cf);
+ void convertFixed12(Vector4i &ci, Vector4f &cf);
+ Float4 convertSigned12(Short4 &ci);
+ void convertSigned12(Vector4f &cf, Vector4i &ci);
+ Float4 convertUnsigned16(UShort4 ci);
+ UShort4 convertFixed16(Float4 &cf, bool saturate = true);
+ void convertFixed16(Vector4i &ci, Vector4f &cf, bool saturate = true);
+ void sRGBtoLinear16_16(Registers &r, Vector4i &c);
+ void sRGBtoLinear12_16(Registers &r, Vector4i &c);
+ void linearToSRGB16_16(Registers &r, Vector4i &c);
+ void linearToSRGB12_16(Registers &r, Vector4i &c);
+ Float4 sRGBtoLinear(const Float4 &x);
+ Float4 linearToSRGB(const Float4 &x);
+
+ // ps_1_x instructions
+ void MOV(Vector4i &dst, Vector4i &src0);
+ void ADD(Vector4i &dst, Vector4i &src0, Vector4i &src1);
+ void SUB(Vector4i &dst, Vector4i &src0, Vector4i &src1);
+ void MAD(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2);
+ void MUL(Vector4i &dst, Vector4i &src0, Vector4i &src1);
+ void DP3(Vector4i &dst, Vector4i &src0, Vector4i &src1);
+ void DP4(Vector4i &dst, Vector4i &src0, Vector4i &src1);
+ void LRP(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2);
+ void TEXCOORD(Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate);
+ void TEXCRD(Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate, bool project);
+ void TEXDP3(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, Vector4i &src);
+ void TEXDP3TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0);
+ void TEXKILL(Int cMask[4], Float4 &u, Float4 &v, Float4 &s);
+ void TEXKILL(Int cMask[4], Vector4i &dst);
+ void TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, bool project);
+ void TEXLD(Registers &r, Vector4i &dst, Vector4i &src, int stage, bool project);
+ void TEXBEM(Registers &r, Vector4i &dst, Vector4i &src, Float4 &u, Float4 &v, Float4 &s, int stage);
+ void TEXBEML(Registers &r, Vector4i &dst, Vector4i &src, Float4 &u, Float4 &v, Float4 &s, int stage);
+ void TEXREG2AR(Registers &r, Vector4i &dst, Vector4i &src0, int stage);
+ void TEXREG2GB(Registers &r, Vector4i &dst, Vector4i &src0, int stage);
+ void TEXREG2RGB(Registers &r, Vector4i &dst, Vector4i &src0, int stage);
+ void TEXM3X2DEPTH(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, Vector4i &src, bool signedScaling);
+ void TEXM3X2PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4i &src0, int component, bool signedScaling);
+ void TEXM3X2TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0, bool signedScaling);
+ void TEXM3X3(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, Vector4i &src0, bool signedScaling);
+ void TEXM3X3PAD(Registers &r, Float4 &u, Float4 &v, Float4 &s, Vector4i &src0, int component, bool signedScaling);
+ void TEXM3X3SPEC(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0, Vector4i &src1);
+ void TEXM3X3TEX(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0, bool singedScaling);
+ void TEXM3X3VSPEC(Registers &r, Vector4i &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4i &src0);
+ void TEXDEPTH(Registers &r);
+ void CND(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2);
+ void CMP(Vector4i &dst, Vector4i &src0, Vector4i &src1, Vector4i &src2);
+ void BEM(Registers &r, Vector4i &dst, Vector4i &src0, Vector4i &src1, int stage);
+
+ // ps_2_x instructions
+ 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 TEXKILL(Int cMask[4], Vector4f &src, unsigned char mask);
+ 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);
+
+ void writeDestination(Registers &r, Vector4i &d, const Dst &dst);
+ Vector4i regi(Registers &r, const Src &src);
+ Vector4f reg(Registers &r, const Src &src, int offset = 0);
+ Vector4f readConstant(Registers &r, const Src &src, int offset = 0);
+ Int relativeAddress(Registers &r, const Shader::Parameter &var);
+ Int4 enableMask(Registers &r, const Shader::Instruction *instruction);
+
+ bool colorUsed();
+ unsigned short shaderVersion() const;
+ bool interpolateZ() const;
+ bool interpolateW() const;
+
+ const PixelShader *const shader;
+
+ private:
+ SamplerCore *sampler[16];
+
+ bool perturbate;
+ bool luminance;
+ bool previousScaling;
+
+ 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;
+ bool isConditionalIf[24 + 24];
+ };
+}
+
+#endif // sw_PixelRoutine_hpp
diff --git a/src/Shader/PixelShader.cpp b/src/Shader/PixelShader.cpp
index 7bf072a..3a43892 100644
--- a/src/Shader/PixelShader.cpp
+++ b/src/Shader/PixelShader.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -15,50 +15,44 @@
namespace sw
{
- PixelShader::PixelShader(const unsigned long *token) : Shader(token)
+ PixelShader::PixelShader(const PixelShader *ps) : Shader()
+ {
+ version = 0x0300;
+ vPosDeclared = false;
+ vFaceDeclared = false;
+ centroid = false;
+
+ if(ps) // Make a copy
+ {
+ for(int i = 0; i < ps->getLength(); i++)
+ {
+ append(new sw::Shader::Instruction(*ps->getInstruction(i)));
+ }
+
+ memcpy(semantic, ps->semantic, sizeof(semantic));
+ vPosDeclared = ps->vPosDeclared;
+ vFaceDeclared = ps->vFaceDeclared;
+ usedSamplers = ps->usedSamplers;
+
+ analyze();
+ }
+ }
+
+ PixelShader::PixelShader(const unsigned long *token) : Shader()
{
parse(token);
+
+ vPosDeclared = false;
+ vFaceDeclared = false;
+ centroid = false;
+
+ analyze();
}
PixelShader::~PixelShader()
{
}
- void PixelShader::parse(const unsigned long *token)
- {
- minorVersion = (unsigned char)(token[0] & 0x000000FF);
- majorVersion = (unsigned char)((token[0] & 0x0000FF00) >> 8);
- shaderType = (ShaderType)((token[0] & 0xFFFF0000) >> 16);
-
- length = validate(token);
- ASSERT(length != 0);
-
- instruction = new Shader::Instruction*[length];
-
- for(int i = 0; i < length; i++)
- {
- while((*token & 0x0000FFFF) == 0x0000FFFE) // Comment token
- {
- int length = (*token & 0x7FFF0000) >> 16;
-
- token += length + 1;
- }
-
- int length = size(*token);
-
- instruction[i] = new Instruction(token, length, majorVersion);
-
- token += length + 1;
- }
-
- analyzeZOverride();
- analyzeTexkill();
- analyzeInterpolants();
- analyzeDirtyConstants();
- analyzeDynamicBranching();
- analyzeSamplers();
- }
-
int PixelShader::validate(const unsigned long *const token)
{
if(!token)
@@ -88,12 +82,12 @@
}
else
{
- ShaderOpcode opcode = (ShaderOpcode)(token[i] & 0x0000FFFF);
+ Shader::Opcode opcode = (Shader::Opcode)(token[i] & 0x0000FFFF);
switch(opcode)
{
- case ShaderOperation::OPCODE_RESERVED0:
- case ShaderOperation::OPCODE_MOVA:
+ case Shader::OPCODE_RESERVED0:
+ case Shader::OPCODE_MOVA:
return 0; // Unsupported operation
default:
instructionCount++;
@@ -112,9 +106,9 @@
return zOverride;
}
- bool PixelShader::containsTexkill() const
+ bool PixelShader::containsKill() const
{
- return texkill;
+ return kill;
}
bool PixelShader::containsCentroid() const
@@ -137,15 +131,27 @@
return semantic[2 + coordinate][component].active();
}
+ void PixelShader::analyze()
+ {
+ analyzeZOverride();
+ analyzeKill();
+ analyzeInterpolants();
+ analyzeDirtyConstants();
+ analyzeDynamicBranching();
+ analyzeSamplers();
+ analyzeCallSites();
+ analyzeDynamicIndexing();
+ }
+
void PixelShader::analyzeZOverride()
{
zOverride = false;
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getOpcode() == Instruction::Operation::OPCODE_TEXM3X2DEPTH ||
- instruction[i]->getOpcode() == Instruction::Operation::OPCODE_TEXDEPTH ||
- instruction[i]->getDestinationParameter().type == Instruction::DestinationParameter::PARAMETER_DEPTHOUT)
+ if(instruction[i]->opcode == Shader::OPCODE_TEXM3X2DEPTH ||
+ instruction[i]->opcode == Shader::OPCODE_TEXDEPTH ||
+ instruction[i]->dst.type == Shader::PARAMETER_DEPTHOUT)
{
zOverride = true;
@@ -154,15 +160,16 @@
}
}
- void PixelShader::analyzeTexkill()
+ void PixelShader::analyzeKill()
{
- texkill = false;
+ kill = false;
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getOpcode() == Instruction::Operation::OPCODE_TEXKILL)
+ if(instruction[i]->opcode == Shader::OPCODE_TEXKILL ||
+ instruction[i]->opcode == Shader::OPCODE_DISCARD)
{
- texkill = true;
+ kill = true;
break;
}
@@ -171,76 +178,72 @@
void PixelShader::analyzeInterpolants()
{
- vPosDeclared = false;
- vFaceDeclared = false;
- centroid = false;
-
if(version < 0x0300)
{
// Set default mapping; disable unused interpolants below
- semantic[0][0] = Semantic(ShaderOperation::USAGE_COLOR, 0);
- semantic[0][1] = Semantic(ShaderOperation::USAGE_COLOR, 0);
- semantic[0][2] = Semantic(ShaderOperation::USAGE_COLOR, 0);
- semantic[0][3] = Semantic(ShaderOperation::USAGE_COLOR, 0);
+ semantic[0][0] = Semantic(Shader::USAGE_COLOR, 0);
+ semantic[0][1] = Semantic(Shader::USAGE_COLOR, 0);
+ semantic[0][2] = Semantic(Shader::USAGE_COLOR, 0);
+ semantic[0][3] = Semantic(Shader::USAGE_COLOR, 0);
- semantic[1][0] = Semantic(ShaderOperation::USAGE_COLOR, 1);
- semantic[1][1] = Semantic(ShaderOperation::USAGE_COLOR, 1);
- semantic[1][2] = Semantic(ShaderOperation::USAGE_COLOR, 1);
- semantic[1][3] = Semantic(ShaderOperation::USAGE_COLOR, 1);
+ semantic[1][0] = Semantic(Shader::USAGE_COLOR, 1);
+ semantic[1][1] = Semantic(Shader::USAGE_COLOR, 1);
+ semantic[1][2] = Semantic(Shader::USAGE_COLOR, 1);
+ semantic[1][3] = Semantic(Shader::USAGE_COLOR, 1);
for(int i = 0; i < 8; i++)
{
- semantic[2 + i][0] = Semantic(ShaderOperation::USAGE_TEXCOORD, i);
- semantic[2 + i][1] = Semantic(ShaderOperation::USAGE_TEXCOORD, i);
- semantic[2 + i][2] = Semantic(ShaderOperation::USAGE_TEXCOORD, i);
- semantic[2 + i][3] = Semantic(ShaderOperation::USAGE_TEXCOORD, i);
+ semantic[2 + i][0] = Semantic(Shader::USAGE_TEXCOORD, i);
+ semantic[2 + i][1] = Semantic(Shader::USAGE_TEXCOORD, i);
+ semantic[2 + i][2] = Semantic(Shader::USAGE_TEXCOORD, i);
+ semantic[2 + i][3] = Semantic(Shader::USAGE_TEXCOORD, i);
}
- Instruction::Operation::SamplerType samplerType[16];
+ Shader::SamplerType samplerType[16];
for(int i = 0; i < 16; i++)
{
- samplerType[i] = Instruction::Operation::SAMPLER_UNKNOWN;
+ samplerType[i] = Shader::SAMPLER_UNKNOWN;
}
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getDestinationParameter().type == Instruction::SourceParameter::PARAMETER_SAMPLER)
+ if(instruction[i]->dst.type == Shader::PARAMETER_SAMPLER)
{
- int sampler = instruction[i]->getDestinationParameter().index;
+ int sampler = instruction[i]->dst.index;
- samplerType[sampler] = instruction[i]->getSamplerType();
+ samplerType[sampler] = instruction[i]->samplerType;
}
}
bool interpolant[10][4] = {false}; // Interpolants in use
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getDestinationParameter().type == Instruction::SourceParameter::PARAMETER_TEXTURE)
+ if(instruction[i]->dst.type == Shader::PARAMETER_TEXTURE)
{
- int index = instruction[i]->getDestinationParameter().index + 2;
- int mask = instruction[i]->getDestinationParameter().mask;
+ int index = instruction[i]->dst.index + 2;
+ int mask = instruction[i]->dst.mask;
- switch(instruction[i]->getOpcode())
+ switch(instruction[i]->opcode)
{
- case Instruction::Operation::OPCODE_TEX:
- case Instruction::Operation::OPCODE_TEXBEM:
- case Instruction::Operation::OPCODE_TEXBEML:
- case Instruction::Operation::OPCODE_TEXCOORD:
- case Instruction::Operation::OPCODE_TEXDP3:
- case Instruction::Operation::OPCODE_TEXDP3TEX:
- case Instruction::Operation::OPCODE_TEXM3X2DEPTH:
- case Instruction::Operation::OPCODE_TEXM3X2PAD:
- case Instruction::Operation::OPCODE_TEXM3X2TEX:
- case Instruction::Operation::OPCODE_TEXM3X3:
- case Instruction::Operation::OPCODE_TEXM3X3PAD:
- case Instruction::Operation::OPCODE_TEXM3X3TEX:
+ case Shader::OPCODE_TEX:
+ case Shader::OPCODE_TEXBEM:
+ case Shader::OPCODE_TEXBEML:
+ case Shader::OPCODE_TEXCOORD:
+ case Shader::OPCODE_TEXDP3:
+ case Shader::OPCODE_TEXDP3TEX:
+ case Shader::OPCODE_TEXM3X2DEPTH:
+ case Shader::OPCODE_TEXM3X2PAD:
+ case Shader::OPCODE_TEXM3X2TEX:
+ case Shader::OPCODE_TEXM3X3:
+ case Shader::OPCODE_TEXM3X3PAD:
+ case Shader::OPCODE_TEXM3X3TEX:
interpolant[index][0] = true;
interpolant[index][1] = true;
interpolant[index][2] = true;
break;
- case Instruction::Operation::OPCODE_TEXKILL:
+ case Shader::OPCODE_TEXKILL:
if(majorVersion < 2)
{
interpolant[index][0] = true;
@@ -255,7 +258,7 @@
interpolant[index][3] = true;
}
break;
- case Instruction::Operation::OPCODE_TEXM3X3VSPEC:
+ case Shader::OPCODE_TEXM3X3VSPEC:
interpolant[index][0] = true;
interpolant[index][1] = true;
interpolant[index][2] = true;
@@ -263,7 +266,7 @@
interpolant[index - 1][3] = true;
interpolant[index - 0][3] = true;
break;
- case Instruction::Operation::OPCODE_DCL:
+ case Shader::OPCODE_DCL:
break; // Ignore
default: // Arithmetic instruction
if(version >= 0x0104)
@@ -275,32 +278,32 @@
for(int argument = 0; argument < 4; argument++)
{
- if(instruction[i]->getSourceParameter(argument).type == Instruction::SourceParameter::PARAMETER_INPUT ||
- instruction[i]->getSourceParameter(argument).type == Instruction::SourceParameter::PARAMETER_TEXTURE)
+ if(instruction[i]->src[argument].type == Shader::PARAMETER_INPUT ||
+ instruction[i]->src[argument].type == Shader::PARAMETER_TEXTURE)
{
- int index = instruction[i]->getSourceParameter(argument).index;
- int swizzle = instruction[i]->getSourceParameter(argument).swizzle;
- int mask = instruction[i]->getDestinationParameter().mask;
+ int index = instruction[i]->src[argument].index;
+ int swizzle = instruction[i]->src[argument].swizzle;
+ int mask = instruction[i]->dst.mask;
- if(instruction[i]->getSourceParameter(argument).type == Instruction::SourceParameter::PARAMETER_TEXTURE)
+ if(instruction[i]->src[argument].type == Shader::PARAMETER_TEXTURE)
{
index += 2;
}
- switch(instruction[i]->getOpcode())
+ switch(instruction[i]->opcode)
{
- case Instruction::Operation::OPCODE_TEX:
- case Instruction::Operation::OPCODE_TEXLDD:
- case Instruction::Operation::OPCODE_TEXLDL:
+ case Shader::OPCODE_TEX:
+ case Shader::OPCODE_TEXLDD:
+ case Shader::OPCODE_TEXLDL:
{
- int sampler = instruction[i]->getSourceParameter(1).index;
+ int sampler = instruction[i]->src[1].index;
switch(samplerType[sampler])
{
- case Instruction::Operation::SAMPLER_UNKNOWN:
+ case Shader::SAMPLER_UNKNOWN:
if(version == 0x0104)
{
- if((instruction[i]->getSourceParameter(0).swizzle & 0x30) == 0x20) // .xyz
+ if((instruction[i]->src[0].swizzle & 0x30) == 0x20) // .xyz
{
interpolant[index][0] = true;
interpolant[index][1] = true;
@@ -318,19 +321,19 @@
ASSERT(false);
}
break;
- case Instruction::Operation::SAMPLER_1D:
+ case Shader::SAMPLER_1D:
interpolant[index][0] = true;
break;
- case Instruction::Operation::SAMPLER_2D:
+ case Shader::SAMPLER_2D:
interpolant[index][0] = true;
interpolant[index][1] = true;
break;
- case Instruction::Operation::SAMPLER_CUBE:
+ case Shader::SAMPLER_CUBE:
interpolant[index][0] = true;
interpolant[index][1] = true;
interpolant[index][2] = true;
break;
- case Instruction::Operation::SAMPLER_VOLUME:
+ case Shader::SAMPLER_VOLUME:
interpolant[index][0] = true;
interpolant[index][1] = true;
interpolant[index][2] = true;
@@ -339,31 +342,31 @@
ASSERT(false);
}
- if(instruction[i]->isBias())
+ if(instruction[i]->bias)
{
interpolant[index][3] = true;
}
- if(instruction[i]->isProject())
+ if(instruction[i]->project)
{
interpolant[index][3] = true;
}
- if(version == 0x0104 && instruction[i]->getOpcode() == Instruction::Operation::OPCODE_TEX)
+ if(version == 0x0104 && instruction[i]->opcode == Shader::OPCODE_TEX)
{
- if(instruction[i]->getSourceParameter(0).modifier == Instruction::SourceParameter::MODIFIER_DZ)
+ if(instruction[i]->src[0].modifier == Shader::MODIFIER_DZ)
{
interpolant[index][2] = true;
}
- if(instruction[i]->getSourceParameter(0).modifier == Instruction::SourceParameter::MODIFIER_DW)
+ if(instruction[i]->src[0].modifier == Shader::MODIFIER_DW)
{
interpolant[index][3] = true;
}
}
}
break;
- case Instruction::Operation::OPCODE_M3X2:
+ case Shader::OPCODE_M3X2:
if(mask & 0x1)
{
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, 0x7);
@@ -383,7 +386,7 @@
}
}
break;
- case Instruction::Operation::OPCODE_M3X3:
+ case Shader::OPCODE_M3X3:
if(mask & 0x1)
{
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, 0x7);
@@ -411,7 +414,7 @@
}
}
break;
- case Instruction::Operation::OPCODE_M3X4:
+ case Shader::OPCODE_M3X4:
if(mask & 0x1)
{
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, 0x7);
@@ -447,7 +450,7 @@
}
}
break;
- case Instruction::Operation::OPCODE_M4X3:
+ case Shader::OPCODE_M4X3:
if(mask & 0x1)
{
interpolant[index][0] |= swizzleContainsComponent(swizzle, 0);
@@ -475,7 +478,7 @@
}
}
break;
- case Instruction::Operation::OPCODE_M4X4:
+ case Shader::OPCODE_M4X4:
if(mask & 0x1)
{
interpolant[index][0] |= swizzleContainsComponent(swizzle, 0);
@@ -511,7 +514,7 @@
}
}
break;
- case Instruction::Operation::OPCODE_CRS:
+ case Shader::OPCODE_CRS:
if(mask & 0x1)
{
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, 0x6);
@@ -536,7 +539,7 @@
interpolant[index][3] |= swizzleContainsComponentMasked(swizzle, 3, 0x3);
}
break;
- case Instruction::Operation::OPCODE_DP2ADD:
+ case Shader::OPCODE_DP2ADD:
if(argument == 0 || argument == 1)
{
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, 0x3);
@@ -552,81 +555,81 @@
interpolant[index][3] |= swizzleContainsComponent(swizzle, 3);
}
break;
- case Instruction::Operation::OPCODE_DP3:
+ case Shader::OPCODE_DP3:
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, 0x7);
interpolant[index][1] |= swizzleContainsComponentMasked(swizzle, 1, 0x7);
interpolant[index][2] |= swizzleContainsComponentMasked(swizzle, 2, 0x7);
interpolant[index][3] |= swizzleContainsComponentMasked(swizzle, 3, 0x7);
break;
- case Instruction::Operation::OPCODE_DP4:
+ case Shader::OPCODE_DP4:
interpolant[index][0] |= swizzleContainsComponent(swizzle, 0);
interpolant[index][1] |= swizzleContainsComponent(swizzle, 1);
interpolant[index][2] |= swizzleContainsComponent(swizzle, 2);
interpolant[index][3] |= swizzleContainsComponent(swizzle, 3);
break;
- case Instruction::Operation::OPCODE_SINCOS:
- case Instruction::Operation::OPCODE_EXP:
- case Instruction::Operation::OPCODE_LOG:
- case Instruction::Operation::OPCODE_POW:
- case Instruction::Operation::OPCODE_RCP:
- case Instruction::Operation::OPCODE_RSQ:
+ case Shader::OPCODE_SINCOS:
+ case Shader::OPCODE_EXP2X:
+ case Shader::OPCODE_LOG2X:
+ case Shader::OPCODE_POWX:
+ case Shader::OPCODE_RCPX:
+ case Shader::OPCODE_RSQX:
interpolant[index][0] |= swizzleContainsComponent(swizzle, 0);
interpolant[index][1] |= swizzleContainsComponent(swizzle, 1);
interpolant[index][2] |= swizzleContainsComponent(swizzle, 2);
interpolant[index][3] |= swizzleContainsComponent(swizzle, 3);
break;
- case Instruction::Operation::OPCODE_NRM:
+ case Shader::OPCODE_NRM3:
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, 0x7 | mask);
interpolant[index][1] |= swizzleContainsComponentMasked(swizzle, 1, 0x7 | mask);
interpolant[index][2] |= swizzleContainsComponentMasked(swizzle, 2, 0x7 | mask);
interpolant[index][3] |= swizzleContainsComponentMasked(swizzle, 3, 0x7 | mask);
break;
- case Instruction::Operation::OPCODE_MOV:
- case Instruction::Operation::OPCODE_ADD:
- case Instruction::Operation::OPCODE_SUB:
- case Instruction::Operation::OPCODE_MUL:
- case Instruction::Operation::OPCODE_MAD:
- case Instruction::Operation::OPCODE_ABS:
- case Instruction::Operation::OPCODE_CMP:
- case Instruction::Operation::OPCODE_CND:
- case Instruction::Operation::OPCODE_FRC:
- case Instruction::Operation::OPCODE_LRP:
- case Instruction::Operation::OPCODE_MAX:
- case Instruction::Operation::OPCODE_MIN:
- case Instruction::Operation::OPCODE_SETP:
- case Instruction::Operation::OPCODE_BREAKC:
- case Instruction::Operation::OPCODE_DSX:
- case Instruction::Operation::OPCODE_DSY:
+ case Shader::OPCODE_MOV:
+ case Shader::OPCODE_ADD:
+ case Shader::OPCODE_SUB:
+ case Shader::OPCODE_MUL:
+ case Shader::OPCODE_MAD:
+ case Shader::OPCODE_ABS:
+ case Shader::OPCODE_CMP0:
+ case Shader::OPCODE_CND:
+ case Shader::OPCODE_FRC:
+ case Shader::OPCODE_LRP:
+ case Shader::OPCODE_MAX:
+ case Shader::OPCODE_MIN:
+ case Shader::OPCODE_CMP:
+ case Shader::OPCODE_BREAKC:
+ case Shader::OPCODE_DFDX:
+ case Shader::OPCODE_DFDY:
interpolant[index][0] |= swizzleContainsComponentMasked(swizzle, 0, mask);
interpolant[index][1] |= swizzleContainsComponentMasked(swizzle, 1, mask);
interpolant[index][2] |= swizzleContainsComponentMasked(swizzle, 2, mask);
interpolant[index][3] |= swizzleContainsComponentMasked(swizzle, 3, mask);
break;
- case Instruction::Operation::OPCODE_TEXCOORD:
+ case Shader::OPCODE_TEXCOORD:
interpolant[index][0] = true;
interpolant[index][1] = true;
interpolant[index][2] = true;
interpolant[index][3] = true;
break;
- case Instruction::Operation::OPCODE_TEXDP3:
- case Instruction::Operation::OPCODE_TEXDP3TEX:
- case Instruction::Operation::OPCODE_TEXM3X2PAD:
- case Instruction::Operation::OPCODE_TEXM3X3PAD:
- case Instruction::Operation::OPCODE_TEXM3X2TEX:
- case Instruction::Operation::OPCODE_TEXM3X3SPEC:
- case Instruction::Operation::OPCODE_TEXM3X3VSPEC:
- case Instruction::Operation::OPCODE_TEXBEM:
- case Instruction::Operation::OPCODE_TEXBEML:
- case Instruction::Operation::OPCODE_TEXM3X2DEPTH:
- case Instruction::Operation::OPCODE_TEXM3X3:
- case Instruction::Operation::OPCODE_TEXM3X3TEX:
+ case Shader::OPCODE_TEXDP3:
+ case Shader::OPCODE_TEXDP3TEX:
+ case Shader::OPCODE_TEXM3X2PAD:
+ case Shader::OPCODE_TEXM3X3PAD:
+ case Shader::OPCODE_TEXM3X2TEX:
+ case Shader::OPCODE_TEXM3X3SPEC:
+ case Shader::OPCODE_TEXM3X3VSPEC:
+ case Shader::OPCODE_TEXBEM:
+ case Shader::OPCODE_TEXBEML:
+ case Shader::OPCODE_TEXM3X2DEPTH:
+ case Shader::OPCODE_TEXM3X3:
+ case Shader::OPCODE_TEXM3X3TEX:
interpolant[index][0] = true;
interpolant[index][1] = true;
interpolant[index][2] = true;
break;
- case Instruction::Operation::OPCODE_TEXREG2AR:
- case Instruction::Operation::OPCODE_TEXREG2GB:
- case Instruction::Operation::OPCODE_TEXREG2RGB:
+ case Shader::OPCODE_TEXREG2AR:
+ case Shader::OPCODE_TEXREG2GB:
+ case Shader::OPCODE_TEXREG2RGB:
break;
default:
// ASSERT(false); // Refine component usage
@@ -652,40 +655,25 @@
}
else // Shader Model 3.0 input declaration; v# indexable
{
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getOpcode() == ShaderOperation::OPCODE_DCL)
+ if(instruction[i]->opcode == Shader::OPCODE_DCL)
{
- if(instruction[i]->getDestinationParameter().type == ShaderParameter::PARAMETER_INPUT)
+ if(instruction[i]->dst.type == Shader::PARAMETER_INPUT)
{
- unsigned char usage = instruction[i]->getUsage();
- unsigned char index = instruction[i]->getUsageIndex();
- unsigned char mask = instruction[i]->getDestinationParameter().mask;
- unsigned char reg = instruction[i]->getDestinationParameter().index;
+ unsigned char usage = instruction[i]->usage;
+ unsigned char index = instruction[i]->usageIndex;
+ unsigned char mask = instruction[i]->dst.mask;
+ unsigned char reg = instruction[i]->dst.index;
- if(mask & 0x01)
- {
- semantic[reg][0] = Semantic(usage, index);
- }
-
- if(mask & 0x02)
- {
- semantic[reg][1] = Semantic(usage, index);
- }
-
- if(mask & 0x04)
- {
- semantic[reg][2] = Semantic(usage, index);
- }
-
- if(mask & 0x08)
- {
- semantic[reg][3] = Semantic(usage, index);
- }
+ if(mask & 0x01) semantic[reg][0] = Semantic(usage, index);
+ if(mask & 0x02) semantic[reg][1] = Semantic(usage, index);
+ if(mask & 0x04) semantic[reg][2] = Semantic(usage, index);
+ if(mask & 0x08) semantic[reg][3] = Semantic(usage, index);
}
- else if(instruction[i]->getDestinationParameter().type == ShaderParameter::PARAMETER_MISCTYPE)
+ else if(instruction[i]->dst.type == Shader::PARAMETER_MISCTYPE)
{
- unsigned char index = instruction[i]->getDestinationParameter().index;
+ unsigned char index = instruction[i]->dst.index;
if(index == 0)
{
@@ -703,19 +691,19 @@
if(version >= 0x0200)
{
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getOpcode() == ShaderOperation::OPCODE_DCL)
+ if(instruction[i]->opcode == Shader::OPCODE_DCL)
{
- bool centroid = instruction[i]->getDestinationParameter().centroid;
- unsigned char reg = instruction[i]->getDestinationParameter().index;
+ bool centroid = instruction[i]->dst.centroid;
+ unsigned char reg = instruction[i]->dst.index;
- switch(instruction[i]->getDestinationParameter().type)
+ switch(instruction[i]->dst.type)
{
- case ShaderParameter::PARAMETER_INPUT:
+ case Shader::PARAMETER_INPUT:
semantic[reg][0].centroid = centroid;
break;
- case ShaderParameter::PARAMETER_TEXTURE:
+ case Shader::PARAMETER_TEXTURE:
semantic[2 + reg][0].centroid = centroid;
break;
}
diff --git a/src/Shader/PixelShader.hpp b/src/Shader/PixelShader.hpp
index a0f9b99..83ca253 100644
--- a/src/Shader/PixelShader.hpp
+++ b/src/Shader/PixelShader.hpp
@@ -1,54 +1,53 @@
-// SwiftShader Software Renderer
-//
-// Copyright(c) 2005-2011 TransGaming Inc.
-//
-// All rights reserved. No part of this software may be copied, distributed, transmitted,
-// transcribed, stored in a retrieval system, translated into any human or computer
-// language by any means, or disclosed to third parties without the explicit written
-// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
-// or implied, including but not limited to any patent rights, are granted to you.
-//
-
-#ifndef sw_PixelShader_hpp
-#define sw_PixelShader_hpp
-
-#include "Shader.hpp"
-
-namespace sw
-{
- class PixelShader : public Shader
- {
- public:
- PixelShader(const unsigned long *token);
-
- virtual ~PixelShader();
-
- static int validate(const unsigned long *const token); // Returns number of instructions if valid
- bool depthOverride() const;
- bool containsTexkill() const;
- bool containsCentroid() const;
- bool usesDiffuse(int component) const;
- bool usesSpecular(int component) const;
- bool usesTexture(int coordinate, int component) const;
-
- Semantic semantic[10][4]; // FIXME: Private
-
- bool vPosDeclared;
- bool vFaceDeclared;
-
- private:
- void parse(const unsigned long *token);
-
- void analyzeZOverride();
- void analyzeTexkill();
- void analyzeInterpolants();
-
- bool zOverride;
- bool texkill;
- bool centroid;
- };
-
- typedef PixelShader::Instruction PixelShaderInstruction;
-}
-
-#endif // sw_PixelShader_hpp
+// SwiftShader Software Renderer
+//
+// Copyright(c) 2005-2012 TransGaming Inc.
+//
+// All rights reserved. No part of this software may be copied, distributed, transmitted,
+// transcribed, stored in a retrieval system, translated into any human or computer
+// language by any means, or disclosed to third parties without the explicit written
+// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
+// or implied, including but not limited to any patent rights, are granted to you.
+//
+
+#ifndef sw_PixelShader_hpp
+#define sw_PixelShader_hpp
+
+#include "Shader.hpp"
+
+namespace sw
+{
+ class PixelShader : public Shader
+ {
+ public:
+ explicit PixelShader(const PixelShader *ps = 0);
+ explicit PixelShader(const unsigned long *token);
+
+ virtual ~PixelShader();
+
+ static int validate(const unsigned long *const token); // Returns number of instructions if valid
+ bool depthOverride() const;
+ bool containsKill() const;
+ bool containsCentroid() const;
+ bool usesDiffuse(int component) const;
+ bool usesSpecular(int component) const;
+ bool usesTexture(int coordinate, int component) const;
+
+ virtual void analyze();
+
+ Semantic semantic[10][4]; // FIXME: Private
+
+ bool vPosDeclared;
+ bool vFaceDeclared;
+
+ private:
+ void analyzeZOverride();
+ void analyzeKill();
+ void analyzeInterpolants();
+
+ bool zOverride;
+ bool kill;
+ bool centroid;
+ };
+}
+
+#endif // sw_PixelShader_hpp
diff --git a/src/Shader/SamplerCore.cpp b/src/Shader/SamplerCore.cpp
index b7055c4..dbd98b8 100644
--- a/src/Shader/SamplerCore.cpp
+++ b/src/Shader/SamplerCore.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -20,7 +20,7 @@
{
}
- void SamplerCore::sampleTexture(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool bias, bool fixed12, bool gradients, bool lodProvided)
+ void SamplerCore::sampleTexture(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias, bool fixed12, bool gradients, bool lodProvided)
{
#if PERF_PROFILE
AddAtomic(Pointer<Long>(&profiler.texOperations), Long(4));
@@ -40,17 +40,17 @@
if(state.textureType == TEXTURE_NULL)
{
- c.r = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- c.g = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- c.b = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ c.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ c.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ c.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
if(fixed12) // FIXME: Convert to fixed12 at higher level, when required
{
- c.a = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ c.w = Short4(0x1000, 0x1000, 0x1000, 0x1000);
}
else
{
- c.a = Short4((short)0xFFFF, (short)0xFFFF, (short)0xFFFF, (short)0xFFFF); // FIXME
+ c.w = Short4((short)0xFFFF, (short)0xFFFF, (short)0xFFFF, (short)0xFFFF); // FIXME
}
}
else
@@ -87,8 +87,8 @@
if(cubeTexture)
{
- uuuu += Float4(0.5f, 0.5f, 0.5f, 0.5f);
- vvvv += Float4(0.5f, 0.5f, 0.5f, 0.5f);
+ uuuu += Float4(0.5f);
+ vvvv += Float4(0.5f);
}
if(!hasFloatTexture())
@@ -97,7 +97,7 @@
}
else
{
- Color4f cf;
+ Vector4f cf;
sampleFloatFilter(texture, cf, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, lodProvided);
@@ -144,40 +144,40 @@
case FORMAT_G8R8:
case FORMAT_G16R16:
case FORMAT_A16B16G16R16:
- if(componentCount < 2) c.g = Short4(0x1000, 0x1000, 0x1000, 0x1000);
- if(componentCount < 3) c.b = Short4(0x1000, 0x1000, 0x1000, 0x1000);
- if(componentCount < 4) c.a = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ if(componentCount < 2) c.y = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ if(componentCount < 3) c.z = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ if(componentCount < 4) c.w = Short4(0x1000, 0x1000, 0x1000, 0x1000);
break;
case FORMAT_A8:
- c.a = c.r;
- c.r = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- c.g = Short4(0x0000, 0x0000, 0x0000, 0x0000);
- c.b = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ c.w = c.x;
+ c.x = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ c.y = Short4(0x0000, 0x0000, 0x0000, 0x0000);
+ c.z = Short4(0x0000, 0x0000, 0x0000, 0x0000);
break;
case FORMAT_L8:
case FORMAT_L16:
- c.g = c.r;
- c.b = c.r;
- c.a = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ c.y = c.x;
+ c.z = c.x;
+ c.w = Short4(0x1000, 0x1000, 0x1000, 0x1000);
break;
case FORMAT_A8L8:
- c.a = c.g;
- c.g = c.r;
- c.b = c.r;
+ c.w = c.y;
+ c.y = c.x;
+ c.z = c.x;
break;
case FORMAT_R32F:
- c.g = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ c.y = Short4(0x1000, 0x1000, 0x1000, 0x1000);
case FORMAT_G32R32F:
- c.b = Short4(0x1000, 0x1000, 0x1000, 0x1000);
- c.a = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ c.z = Short4(0x1000, 0x1000, 0x1000, 0x1000);
+ c.w = Short4(0x1000, 0x1000, 0x1000, 0x1000);
case FORMAT_A32B32G32R32F:
break;
case FORMAT_D32F_LOCKABLE:
case FORMAT_D32F_TEXTURE:
case FORMAT_D32F_SHADOW:
- c.g = c.r;
- c.b = c.r;
- c.a = c.r;
+ c.y = c.x;
+ c.z = c.x;
+ c.w = c.x;
break;
default:
ASSERT(false);
@@ -186,7 +186,7 @@
}
}
- void SamplerCore::sampleTexture(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool bias, bool gradients, bool lodProvided)
+ void SamplerCore::sampleTexture(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided)
{
#if PERF_PROFILE
AddAtomic(Pointer<Long>(&profiler.texOperations), Long(4));
@@ -202,10 +202,10 @@
if(state.textureType == TEXTURE_NULL)
{
- c.r = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- c.g = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- c.b = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- c.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ c.x = Float4(0.0f);
+ c.y = Float4(0.0f);
+ c.z = Float4(0.0f);
+ c.w = Float4(1.0f);
}
else
{
@@ -247,15 +247,15 @@
if(cubeTexture)
{
- uuuu += Float4(0.5f, 0.5f, 0.5f, 0.5f);
- vvvv += Float4(0.5f, 0.5f, 0.5f, 0.5f);
+ uuuu += Float4(0.5f);
+ vvvv += Float4(0.5f);
}
sampleFloatFilter(texture, c, uuuu, vvvv, wwww, lod, anisotropy, uDelta, vDelta, face, lodProvided);
}
else
{
- Color4i ci;
+ Vector4i ci;
sampleTexture(texture, ci, u, v, w, q, dsx, dsy, bias, false, gradients, lodProvided);
@@ -298,40 +298,40 @@
case FORMAT_G8R8:
case FORMAT_G16R16:
case FORMAT_A16B16G16R16:
- if(componentCount < 2) c.g = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- if(componentCount < 3) c.b = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- if(componentCount < 4) c.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ if(componentCount < 2) c.y = Float4(1.0f);
+ if(componentCount < 3) c.z = Float4(1.0f);
+ if(componentCount < 4) c.w = Float4(1.0f);
break;
case FORMAT_A8:
- c.a = c.r;
- c.r = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- c.g = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- c.b = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ c.w = c.x;
+ c.x = Float4(0.0f);
+ c.y = Float4(0.0f);
+ c.z = Float4(0.0f);
break;
case FORMAT_L8:
case FORMAT_L16:
- c.g = c.r;
- c.b = c.r;
- c.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ c.y = c.x;
+ c.z = c.x;
+ c.w = Float4(1.0f);
break;
case FORMAT_A8L8:
- c.a = c.g;
- c.g = c.r;
- c.b = c.r;
+ c.w = c.y;
+ c.y = c.x;
+ c.z = c.x;
break;
case FORMAT_R32F:
- c.g = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ c.y = Float4(1.0f);
case FORMAT_G32R32F:
- c.b = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- c.a = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ c.z = Float4(1.0f);
+ c.w = Float4(1.0f);
case FORMAT_A32B32G32R32F:
break;
case FORMAT_D32F_LOCKABLE:
case FORMAT_D32F_TEXTURE:
case FORMAT_D32F_SHADOW:
- c.g = c.r;
- c.b = c.r;
- c.a = c.r;
+ c.y = c.x;
+ c.z = c.x;
+ c.w = c.x;
break;
default:
ASSERT(false);
@@ -377,7 +377,7 @@
return uvw;
}
- void SamplerCore::sampleFilter(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided)
+ void SamplerCore::sampleFilter(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided)
{
bool volumeTexture = state.textureType == TEXTURE_3D;
@@ -385,7 +385,7 @@
if(state.mipmapFilter > MIPMAP_POINT)
{
- Color4i cc;
+ Vector4i cc;
sampleAniso(texture, cc, u, v, w, lod, anisotropy, uDelta, vDelta, face, true, lodProvided);
@@ -394,28 +394,28 @@
UShort4 utri = UShort4(Float4(lod)); // FIXME: Optimize
Short4 stri = utri >> 1; // FIXME: Optimize
- if(hasUnsignedTextureComponent(0)) cc.r = MulHigh(As<UShort4>(cc.r), utri); else cc.r = MulHigh(cc.r, stri);
- if(hasUnsignedTextureComponent(1)) cc.g = MulHigh(As<UShort4>(cc.g), utri); else cc.g = MulHigh(cc.g, stri);
- if(hasUnsignedTextureComponent(2)) cc.b = MulHigh(As<UShort4>(cc.b), utri); else cc.b = MulHigh(cc.b, stri);
- if(hasUnsignedTextureComponent(3)) cc.a = MulHigh(As<UShort4>(cc.a), utri); else cc.a = MulHigh(cc.a, stri);
+ if(hasUnsignedTextureComponent(0)) cc.x = MulHigh(As<UShort4>(cc.x), utri); else cc.x = MulHigh(cc.x, stri);
+ if(hasUnsignedTextureComponent(1)) cc.y = MulHigh(As<UShort4>(cc.y), utri); else cc.y = MulHigh(cc.y, stri);
+ if(hasUnsignedTextureComponent(2)) cc.z = MulHigh(As<UShort4>(cc.z), utri); else cc.z = MulHigh(cc.z, stri);
+ if(hasUnsignedTextureComponent(3)) cc.w = MulHigh(As<UShort4>(cc.w), utri); else cc.w = MulHigh(cc.w, stri);
utri = ~utri;
stri = Short4(0x7FFF, 0x7FFF, 0x7FFF, 0x7FFF) - stri;
- if(hasUnsignedTextureComponent(0)) c.r = MulHigh(As<UShort4>(c.r), utri); else c.r = MulHigh(c.r, stri);
- if(hasUnsignedTextureComponent(1)) c.g = MulHigh(As<UShort4>(c.g), utri); else c.g = MulHigh(c.g, stri);
- if(hasUnsignedTextureComponent(2)) c.b = MulHigh(As<UShort4>(c.b), utri); else c.b = MulHigh(c.b, stri);
- if(hasUnsignedTextureComponent(3)) c.a = MulHigh(As<UShort4>(c.a), utri); else c.a = MulHigh(c.a, stri);
+ if(hasUnsignedTextureComponent(0)) c.x = MulHigh(As<UShort4>(c.x), utri); else c.x = MulHigh(c.x, stri);
+ if(hasUnsignedTextureComponent(1)) c.y = MulHigh(As<UShort4>(c.y), utri); else c.y = MulHigh(c.y, stri);
+ if(hasUnsignedTextureComponent(2)) c.z = MulHigh(As<UShort4>(c.z), utri); else c.z = MulHigh(c.z, stri);
+ if(hasUnsignedTextureComponent(3)) c.w = MulHigh(As<UShort4>(c.w), utri); else c.w = MulHigh(c.w, stri);
- c.r += cc.r;
- c.g += cc.g;
- c.b += cc.b;
- c.a += cc.a;
+ c.x += cc.x;
+ c.y += cc.y;
+ c.z += cc.z;
+ c.w += cc.w;
- if(!hasUnsignedTextureComponent(0)) c.r += c.r;
- if(!hasUnsignedTextureComponent(1)) c.g += c.g;
- if(!hasUnsignedTextureComponent(2)) c.b += c.b;
- if(!hasUnsignedTextureComponent(3)) c.a += c.a;
+ if(!hasUnsignedTextureComponent(0)) c.x += c.x;
+ if(!hasUnsignedTextureComponent(1)) c.y += c.y;
+ if(!hasUnsignedTextureComponent(2)) c.z += c.z;
+ if(!hasUnsignedTextureComponent(3)) c.w += c.w;
}
Short4 borderMask;
@@ -468,14 +468,14 @@
{
Short4 b;
- c.r = borderMask & c.r | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[0])) >> (hasUnsignedTextureComponent(0) ? 0 : 1));
- c.g = borderMask & c.g | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[1])) >> (hasUnsignedTextureComponent(1) ? 0 : 1));
- c.b = borderMask & c.b | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[2])) >> (hasUnsignedTextureComponent(2) ? 0 : 1));
- c.a = borderMask & c.a | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[3])) >> (hasUnsignedTextureComponent(3) ? 0 : 1));
+ c.x = borderMask & c.x | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[0])) >> (hasUnsignedTextureComponent(0) ? 0 : 1));
+ c.y = borderMask & c.y | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[1])) >> (hasUnsignedTextureComponent(1) ? 0 : 1));
+ c.z = borderMask & c.z | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[2])) >> (hasUnsignedTextureComponent(2) ? 0 : 1));
+ c.w = borderMask & c.w | ~borderMask & (*Pointer<Short4>(texture + OFFSET(Texture,borderColor4[3])) >> (hasUnsignedTextureComponent(3) ? 0 : 1));
}
}
- void SamplerCore::sampleAniso(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided)
+ void SamplerCore::sampleAniso(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided)
{
if(state.textureFilter != FILTER_ANISOTROPIC || lodProvided)
{
@@ -485,12 +485,12 @@
{
Int a = RoundInt(anisotropy);
- Color4i cSum;
+ Vector4i cSum;
- cSum.r = Short4(0, 0, 0, 0);
- cSum.g = Short4(0, 0, 0, 0);
- cSum.b = Short4(0, 0, 0, 0);
- cSum.a = Short4(0, 0, 0, 0);
+ cSum.x = Short4(0, 0, 0, 0);
+ cSum.y = Short4(0, 0, 0, 0);
+ cSum.z = Short4(0, 0, 0, 0);
+ cSum.w = Short4(0, 0, 0, 0);
Float4 A = *Pointer<Float4>(constants + OFFSET(Constants,uvWeight) + 16 * a);
Float4 B = *Pointer<Float4>(constants + OFFSET(Constants,uvStart) + 16 * a);
@@ -515,23 +515,23 @@
u0 += du;
v0 += dv;
- if(hasUnsignedTextureComponent(0)) cSum.r += As<Short4>(MulHigh(As<UShort4>(c.r), cw)); else cSum.r += MulHigh(c.r, sw);
- if(hasUnsignedTextureComponent(1)) cSum.g += As<Short4>(MulHigh(As<UShort4>(c.g), cw)); else cSum.g += MulHigh(c.g, sw);
- if(hasUnsignedTextureComponent(2)) cSum.b += As<Short4>(MulHigh(As<UShort4>(c.b), cw)); else cSum.b += MulHigh(c.b, sw);
- if(hasUnsignedTextureComponent(3)) cSum.a += As<Short4>(MulHigh(As<UShort4>(c.a), cw)); else cSum.a += MulHigh(c.a, sw);
+ if(hasUnsignedTextureComponent(0)) cSum.x += As<Short4>(MulHigh(As<UShort4>(c.x), cw)); else cSum.x += MulHigh(c.x, sw);
+ if(hasUnsignedTextureComponent(1)) cSum.y += As<Short4>(MulHigh(As<UShort4>(c.y), cw)); else cSum.y += MulHigh(c.y, sw);
+ if(hasUnsignedTextureComponent(2)) cSum.z += As<Short4>(MulHigh(As<UShort4>(c.z), cw)); else cSum.z += MulHigh(c.z, sw);
+ if(hasUnsignedTextureComponent(3)) cSum.w += As<Short4>(MulHigh(As<UShort4>(c.w), cw)); else cSum.w += MulHigh(c.w, sw);
i++;
}
Until(i >= a)
- if(hasUnsignedTextureComponent(0)) c.r = cSum.r; else c.r = AddSat(cSum.r, cSum.r);
- if(hasUnsignedTextureComponent(1)) c.g = cSum.g; else c.g = AddSat(cSum.g, cSum.g);
- if(hasUnsignedTextureComponent(2)) c.b = cSum.b; else c.b = AddSat(cSum.b, cSum.b);
- if(hasUnsignedTextureComponent(3)) c.a = cSum.a; else c.a = AddSat(cSum.a, cSum.a);
+ if(hasUnsignedTextureComponent(0)) c.x = cSum.x; else c.x = AddSat(cSum.x, cSum.x);
+ if(hasUnsignedTextureComponent(1)) c.y = cSum.y; else c.y = AddSat(cSum.y, cSum.y);
+ if(hasUnsignedTextureComponent(2)) c.z = cSum.z; else c.z = AddSat(cSum.z, cSum.z);
+ if(hasUnsignedTextureComponent(3)) c.w = cSum.w; else c.w = AddSat(cSum.w, cSum.w);
}
}
- void SamplerCore::sampleQuad(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD)
+ void SamplerCore::sampleQuad(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD)
{
if(state.textureType != TEXTURE_3D)
{
@@ -543,7 +543,7 @@
}
}
- void SamplerCore::sampleQuad2D(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float &lod, Int face[4], bool secondLOD)
+ void SamplerCore::sampleQuad2D(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float &lod, Int face[4], bool secondLOD)
{
int componentCount = textureComponentCount();
bool gather = state.textureFilter == FILTER_GATHER;
@@ -565,10 +565,10 @@
}
else
{
- Color4i c0;
- Color4i c1;
- Color4i c2;
- Color4i c3;
+ Vector4i c0;
+ Vector4i c1;
+ Vector4i c2;
+ Vector4i c3;
Short4 uuuu0 = offsetSample(uuuu, mipmap, OFFSET(Mipmap,uHalf), (AddressingMode)state.addressingModeU == ADDRESSING_WRAP, gather ? 0 : -1);
Short4 vvvv0 = offsetSample(vvvv, mipmap, OFFSET(Mipmap,vHalf), (AddressingMode)state.addressingModeV == ADDRESSING_WRAP, gather ? 0 : -1);
@@ -624,29 +624,29 @@
{
if(has16bitTexture() && hasUnsignedTextureComponent(0))
{
- c0.r = As<UShort4>(c0.r) - MulHigh(As<UShort4>(c0.r), f0u) + MulHigh(As<UShort4>(c1.r), f0u);
- c2.r = As<UShort4>(c2.r) - MulHigh(As<UShort4>(c2.r), f0u) + MulHigh(As<UShort4>(c3.r), f0u);
- c.r = As<UShort4>(c0.r) - MulHigh(As<UShort4>(c0.r), f0v) + MulHigh(As<UShort4>(c2.r), f0v);
+ c0.x = As<UShort4>(c0.x) - MulHigh(As<UShort4>(c0.x), f0u) + MulHigh(As<UShort4>(c1.x), f0u);
+ c2.x = As<UShort4>(c2.x) - MulHigh(As<UShort4>(c2.x), f0u) + MulHigh(As<UShort4>(c3.x), f0u);
+ c.x = As<UShort4>(c0.x) - MulHigh(As<UShort4>(c0.x), f0v) + MulHigh(As<UShort4>(c2.x), f0v);
}
else
{
if(hasUnsignedTextureComponent(0))
{
- c0.r = MulHigh(As<UShort4>(c0.r), As<UShort4>(f1u1v));
- c1.r = MulHigh(As<UShort4>(c1.r), As<UShort4>(f0u1v));
- c2.r = MulHigh(As<UShort4>(c2.r), As<UShort4>(f1u0v));
- c3.r = MulHigh(As<UShort4>(c3.r), As<UShort4>(f0u0v));
+ c0.x = MulHigh(As<UShort4>(c0.x), As<UShort4>(f1u1v));
+ c1.x = MulHigh(As<UShort4>(c1.x), As<UShort4>(f0u1v));
+ c2.x = MulHigh(As<UShort4>(c2.x), As<UShort4>(f1u0v));
+ c3.x = MulHigh(As<UShort4>(c3.x), As<UShort4>(f0u0v));
}
else
{
- c0.r = MulHigh(c0.r, f1u1vs);
- c1.r = MulHigh(c1.r, f0u1vs);
- c2.r = MulHigh(c2.r, f1u0vs);
- c3.r = MulHigh(c3.r, f0u0vs);
+ c0.x = MulHigh(c0.x, f1u1vs);
+ c1.x = MulHigh(c1.x, f0u1vs);
+ c2.x = MulHigh(c2.x, f1u0vs);
+ c3.x = MulHigh(c3.x, f0u0vs);
}
- c.r = (c0.r + c1.r) + (c2.r + c3.r);
- if(!hasUnsignedTextureComponent(0)) c.r = AddSat(c.r, c.r); // Correct for signed fractions
+ c.x = (c0.x + c1.x) + (c2.x + c3.x);
+ if(!hasUnsignedTextureComponent(0)) c.x = AddSat(c.x, c.x); // Correct for signed fractions
}
}
@@ -654,29 +654,29 @@
{
if(has16bitTexture() && hasUnsignedTextureComponent(1))
{
- c0.g = As<UShort4>(c0.g) - MulHigh(As<UShort4>(c0.g), f0u) + MulHigh(As<UShort4>(c1.g), f0u);
- c2.g = As<UShort4>(c2.g) - MulHigh(As<UShort4>(c2.g), f0u) + MulHigh(As<UShort4>(c3.g), f0u);
- c.g = As<UShort4>(c0.g) - MulHigh(As<UShort4>(c0.g), f0v) + MulHigh(As<UShort4>(c2.g), f0v);
+ c0.y = As<UShort4>(c0.y) - MulHigh(As<UShort4>(c0.y), f0u) + MulHigh(As<UShort4>(c1.y), f0u);
+ c2.y = As<UShort4>(c2.y) - MulHigh(As<UShort4>(c2.y), f0u) + MulHigh(As<UShort4>(c3.y), f0u);
+ c.y = As<UShort4>(c0.y) - MulHigh(As<UShort4>(c0.y), f0v) + MulHigh(As<UShort4>(c2.y), f0v);
}
else
{
if(hasUnsignedTextureComponent(1))
{
- c0.g = MulHigh(As<UShort4>(c0.g), As<UShort4>(f1u1v));
- c1.g = MulHigh(As<UShort4>(c1.g), As<UShort4>(f0u1v));
- c2.g = MulHigh(As<UShort4>(c2.g), As<UShort4>(f1u0v));
- c3.g = MulHigh(As<UShort4>(c3.g), As<UShort4>(f0u0v));
+ c0.y = MulHigh(As<UShort4>(c0.y), As<UShort4>(f1u1v));
+ c1.y = MulHigh(As<UShort4>(c1.y), As<UShort4>(f0u1v));
+ c2.y = MulHigh(As<UShort4>(c2.y), As<UShort4>(f1u0v));
+ c3.y = MulHigh(As<UShort4>(c3.y), As<UShort4>(f0u0v));
}
else
{
- c0.g = MulHigh(c0.g, f1u1vs);
- c1.g = MulHigh(c1.g, f0u1vs);
- c2.g = MulHigh(c2.g, f1u0vs);
- c3.g = MulHigh(c3.g, f0u0vs);
+ c0.y = MulHigh(c0.y, f1u1vs);
+ c1.y = MulHigh(c1.y, f0u1vs);
+ c2.y = MulHigh(c2.y, f1u0vs);
+ c3.y = MulHigh(c3.y, f0u0vs);
}
- c.g = (c0.g + c1.g) + (c2.g + c3.g);
- if(!hasUnsignedTextureComponent(1)) c.g = AddSat(c.g, c.g); // Correct for signed fractions
+ c.y = (c0.y + c1.y) + (c2.y + c3.y);
+ if(!hasUnsignedTextureComponent(1)) c.y = AddSat(c.y, c.y); // Correct for signed fractions
}
}
@@ -684,29 +684,29 @@
{
if(has16bitTexture() && hasUnsignedTextureComponent(2))
{
- c0.b = As<UShort4>(c0.b) - MulHigh(As<UShort4>(c0.b), f0u) + MulHigh(As<UShort4>(c1.b), f0u);
- c2.b = As<UShort4>(c2.b) - MulHigh(As<UShort4>(c2.b), f0u) + MulHigh(As<UShort4>(c3.b), f0u);
- c.b = As<UShort4>(c0.b) - MulHigh(As<UShort4>(c0.b), f0v) + MulHigh(As<UShort4>(c2.b), f0v);
+ c0.z = As<UShort4>(c0.z) - MulHigh(As<UShort4>(c0.z), f0u) + MulHigh(As<UShort4>(c1.z), f0u);
+ c2.z = As<UShort4>(c2.z) - MulHigh(As<UShort4>(c2.z), f0u) + MulHigh(As<UShort4>(c3.z), f0u);
+ c.z = As<UShort4>(c0.z) - MulHigh(As<UShort4>(c0.z), f0v) + MulHigh(As<UShort4>(c2.z), f0v);
}
else
{
if(hasUnsignedTextureComponent(2))
{
- c0.b = MulHigh(As<UShort4>(c0.b), As<UShort4>(f1u1v));
- c1.b = MulHigh(As<UShort4>(c1.b), As<UShort4>(f0u1v));
- c2.b = MulHigh(As<UShort4>(c2.b), As<UShort4>(f1u0v));
- c3.b = MulHigh(As<UShort4>(c3.b), As<UShort4>(f0u0v));
+ c0.z = MulHigh(As<UShort4>(c0.z), As<UShort4>(f1u1v));
+ c1.z = MulHigh(As<UShort4>(c1.z), As<UShort4>(f0u1v));
+ c2.z = MulHigh(As<UShort4>(c2.z), As<UShort4>(f1u0v));
+ c3.z = MulHigh(As<UShort4>(c3.z), As<UShort4>(f0u0v));
}
else
{
- c0.b = MulHigh(c0.b, f1u1vs);
- c1.b = MulHigh(c1.b, f0u1vs);
- c2.b = MulHigh(c2.b, f1u0vs);
- c3.b = MulHigh(c3.b, f0u0vs);
+ c0.z = MulHigh(c0.z, f1u1vs);
+ c1.z = MulHigh(c1.z, f0u1vs);
+ c2.z = MulHigh(c2.z, f1u0vs);
+ c3.z = MulHigh(c3.z, f0u0vs);
}
- c.b = (c0.b + c1.b) + (c2.b + c3.b);
- if(!hasUnsignedTextureComponent(2)) c.b = AddSat(c.b, c.b); // Correct for signed fractions
+ c.z = (c0.z + c1.z) + (c2.z + c3.z);
+ if(!hasUnsignedTextureComponent(2)) c.z = AddSat(c.z, c.z); // Correct for signed fractions
}
}
@@ -714,43 +714,43 @@
{
if(has16bitTexture() && hasUnsignedTextureComponent(3))
{
- c0.a = As<UShort4>(c0.a) - MulHigh(As<UShort4>(c0.a), f0u) + MulHigh(As<UShort4>(c1.a), f0u);
- c2.a = As<UShort4>(c2.a) - MulHigh(As<UShort4>(c2.a), f0u) + MulHigh(As<UShort4>(c3.a), f0u);
- c.a = As<UShort4>(c0.a) - MulHigh(As<UShort4>(c0.a), f0v) + MulHigh(As<UShort4>(c2.a), f0v);
+ c0.w = As<UShort4>(c0.w) - MulHigh(As<UShort4>(c0.w), f0u) + MulHigh(As<UShort4>(c1.w), f0u);
+ c2.w = As<UShort4>(c2.w) - MulHigh(As<UShort4>(c2.w), f0u) + MulHigh(As<UShort4>(c3.w), f0u);
+ c.w = As<UShort4>(c0.w) - MulHigh(As<UShort4>(c0.w), f0v) + MulHigh(As<UShort4>(c2.w), f0v);
}
else
{
if(hasUnsignedTextureComponent(3))
{
- c0.a = MulHigh(As<UShort4>(c0.a), As<UShort4>(f1u1v));
- c1.a = MulHigh(As<UShort4>(c1.a), As<UShort4>(f0u1v));
- c2.a = MulHigh(As<UShort4>(c2.a), As<UShort4>(f1u0v));
- c3.a = MulHigh(As<UShort4>(c3.a), As<UShort4>(f0u0v));
+ c0.w = MulHigh(As<UShort4>(c0.w), As<UShort4>(f1u1v));
+ c1.w = MulHigh(As<UShort4>(c1.w), As<UShort4>(f0u1v));
+ c2.w = MulHigh(As<UShort4>(c2.w), As<UShort4>(f1u0v));
+ c3.w = MulHigh(As<UShort4>(c3.w), As<UShort4>(f0u0v));
}
else
{
- c0.a = MulHigh(c0.a, f1u1vs);
- c1.a = MulHigh(c1.a, f0u1vs);
- c2.a = MulHigh(c2.a, f1u0vs);
- c3.a = MulHigh(c3.a, f0u0vs);
+ c0.w = MulHigh(c0.w, f1u1vs);
+ c1.w = MulHigh(c1.w, f0u1vs);
+ c2.w = MulHigh(c2.w, f1u0vs);
+ c3.w = MulHigh(c3.w, f0u0vs);
}
- c.a = (c0.a + c1.a) + (c2.a + c3.a);
- if(!hasUnsignedTextureComponent(3)) c.a = AddSat(c.a, c.a); // Correct for signed fractions
+ c.w = (c0.w + c1.w) + (c2.w + c3.w);
+ if(!hasUnsignedTextureComponent(3)) c.w = AddSat(c.w, c.w); // Correct for signed fractions
}
}
}
else
{
- c.r = c1.r;
- c.g = c2.r;
- c.b = c3.r;
- c.a = c0.r;
+ c.x = c1.x;
+ c.y = c2.x;
+ c.z = c3.x;
+ c.w = c0.x;
}
}
}
- void SamplerCore::sample3D(Pointer<Byte> &texture, Color4i &c_, Float4 &u_, Float4 &v_, Float4 &w_, Float &lod, bool secondLOD)
+ void SamplerCore::sample3D(Pointer<Byte> &texture, Vector4i &c_, Float4 &u_, Float4 &v_, Float4 &w_, Float &lod, bool secondLOD)
{
int componentCount = textureComponentCount();
@@ -774,7 +774,7 @@
}
else
{
- Color4i c[2][2][2];
+ Vector4i c[2][2][2];
Short4 u[2][2][2];
Short4 v[2][2][2];
@@ -863,36 +863,36 @@
{
sampleTexel(c[i][j][k], u[i][j][k], v[i][j][k], s[i][j][k], mipmap, buffer);
- if(componentCount >= 1) if(hasUnsignedTextureComponent(0)) c[i][j][k].r = MulHigh(As<UShort4>(c[i][j][k].r), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].r = MulHigh(c[i][j][k].r, fs[1 - i][1 - j][1 - k]);
- if(componentCount >= 2) if(hasUnsignedTextureComponent(1)) c[i][j][k].g = MulHigh(As<UShort4>(c[i][j][k].g), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].g = MulHigh(c[i][j][k].g, fs[1 - i][1 - j][1 - k]);
- if(componentCount >= 3) if(hasUnsignedTextureComponent(2)) c[i][j][k].b = MulHigh(As<UShort4>(c[i][j][k].b), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].b = MulHigh(c[i][j][k].b, fs[1 - i][1 - j][1 - k]);
- if(componentCount >= 4) if(hasUnsignedTextureComponent(3)) c[i][j][k].a = MulHigh(As<UShort4>(c[i][j][k].a), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].a = MulHigh(c[i][j][k].a, fs[1 - i][1 - j][1 - k]);
+ if(componentCount >= 1) if(hasUnsignedTextureComponent(0)) c[i][j][k].x = MulHigh(As<UShort4>(c[i][j][k].x), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].x = MulHigh(c[i][j][k].x, fs[1 - i][1 - j][1 - k]);
+ if(componentCount >= 2) if(hasUnsignedTextureComponent(1)) c[i][j][k].y = MulHigh(As<UShort4>(c[i][j][k].y), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].y = MulHigh(c[i][j][k].y, fs[1 - i][1 - j][1 - k]);
+ if(componentCount >= 3) if(hasUnsignedTextureComponent(2)) c[i][j][k].z = MulHigh(As<UShort4>(c[i][j][k].z), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].z = MulHigh(c[i][j][k].z, fs[1 - i][1 - j][1 - k]);
+ if(componentCount >= 4) if(hasUnsignedTextureComponent(3)) c[i][j][k].w = MulHigh(As<UShort4>(c[i][j][k].w), As<UShort4>(f[1 - i][1 - j][1 - k])); else c[i][j][k].w = MulHigh(c[i][j][k].w, fs[1 - i][1 - j][1 - k]);
if(i != 0 || j != 0 || k != 0)
{
- if(componentCount >= 1) c[0][0][0].r += c[i][j][k].r;
- if(componentCount >= 2) c[0][0][0].g += c[i][j][k].g;
- if(componentCount >= 3) c[0][0][0].b += c[i][j][k].b;
- if(componentCount >= 4) c[0][0][0].a += c[i][j][k].a;
+ if(componentCount >= 1) c[0][0][0].x += c[i][j][k].x;
+ if(componentCount >= 2) c[0][0][0].y += c[i][j][k].y;
+ if(componentCount >= 3) c[0][0][0].z += c[i][j][k].z;
+ if(componentCount >= 4) c[0][0][0].w += c[i][j][k].w;
}
}
}
}
- if(componentCount >= 1) c_.r = c[0][0][0].r;
- if(componentCount >= 2) c_.g = c[0][0][0].g;
- if(componentCount >= 3) c_.b = c[0][0][0].b;
- if(componentCount >= 4) c_.a = c[0][0][0].a;
+ if(componentCount >= 1) c_.x = c[0][0][0].x;
+ if(componentCount >= 2) c_.y = c[0][0][0].y;
+ if(componentCount >= 3) c_.z = c[0][0][0].z;
+ if(componentCount >= 4) c_.w = c[0][0][0].w;
// Correct for signed fractions
- if(componentCount >= 1) if(!hasUnsignedTextureComponent(0)) c_.r = AddSat(c_.r, c_.r);
- if(componentCount >= 2) if(!hasUnsignedTextureComponent(1)) c_.g = AddSat(c_.g, c_.g);
- if(componentCount >= 3) if(!hasUnsignedTextureComponent(2)) c_.b = AddSat(c_.b, c_.b);
- if(componentCount >= 4) if(!hasUnsignedTextureComponent(3)) c_.a = AddSat(c_.a, c_.a);
+ if(componentCount >= 1) if(!hasUnsignedTextureComponent(0)) c_.x = AddSat(c_.x, c_.x);
+ if(componentCount >= 2) if(!hasUnsignedTextureComponent(1)) c_.y = AddSat(c_.y, c_.y);
+ if(componentCount >= 3) if(!hasUnsignedTextureComponent(2)) c_.z = AddSat(c_.z, c_.z);
+ if(componentCount >= 4) if(!hasUnsignedTextureComponent(3)) c_.w = AddSat(c_.w, c_.w);
}
}
- void SamplerCore::sampleFloatFilter(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided)
+ void SamplerCore::sampleFloatFilter(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided)
{
bool volumeTexture = state.textureType == TEXTURE_3D;
@@ -900,16 +900,16 @@
if(state.mipmapFilter > MIPMAP_POINT)
{
- Color4f cc;
+ Vector4f cc;
sampleFloatAniso(texture, cc, u, v, w, lod, anisotropy, uDelta, vDelta, face, true, lodProvided);
- Float4 lod4 = Float4(Fraction(lod));
+ Float4 lod4 = Float4(Frac(lod));
- c.r = (cc.r - c.r) * lod4 + c.r;
- c.g = (cc.g - c.g) * lod4 + c.g;
- c.b = (cc.b - c.b) * lod4 + c.b;
- c.a = (cc.a - c.a) * lod4 + c.a;
+ c.x = (cc.x - c.x) * lod4 + c.x;
+ c.y = (cc.y - c.y) * lod4 + c.y;
+ c.z = (cc.z - c.z) * lod4 + c.z;
+ c.w = (cc.w - c.w) * lod4 + c.w;
}
Int4 borderMask;
@@ -962,14 +962,14 @@
{
Int4 b;
- c.r = As<Float4>(borderMask & As<Int4>(c.r) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[0])));
- c.g = As<Float4>(borderMask & As<Int4>(c.g) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[1])));
- c.b = As<Float4>(borderMask & As<Int4>(c.b) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[2])));
- c.a = As<Float4>(borderMask & As<Int4>(c.a) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[3])));
+ c.x = As<Float4>(borderMask & As<Int4>(c.x) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[0])));
+ c.y = As<Float4>(borderMask & As<Int4>(c.y) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[1])));
+ c.z = As<Float4>(borderMask & As<Int4>(c.z) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[2])));
+ c.w = As<Float4>(borderMask & As<Int4>(c.w) | ~borderMask & *Pointer<Int4>(texture + OFFSET(Texture,borderColorF[3])));
}
}
- void SamplerCore::sampleFloatAniso(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided)
+ void SamplerCore::sampleFloatAniso(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided)
{
if(state.textureFilter != FILTER_ANISOTROPIC || lodProvided)
{
@@ -979,12 +979,12 @@
{
Int a = RoundInt(anisotropy);
- Color4f cSum;
+ Vector4f cSum;
- cSum.r = Float4(0, 0, 0, 0);
- cSum.g = Float4(0, 0, 0, 0);
- cSum.b = Float4(0, 0, 0, 0);
- cSum.a = Float4(0, 0, 0, 0);
+ cSum.x = Float4(0.0f);
+ cSum.y = Float4(0.0f);
+ cSum.z = Float4(0.0f);
+ cSum.w = Float4(0.0f);
Float4 A = *Pointer<Float4>(constants + OFFSET(Constants,uvWeight) + 16 * a);
Float4 B = *Pointer<Float4>(constants + OFFSET(Constants,uvStart) + 16 * a);
@@ -1007,23 +1007,23 @@
u0 += du;
v0 += dv;
- cSum.r += c.r * A;
- cSum.g += c.g * A;
- cSum.b += c.b * A;
- cSum.a += c.a * A;
+ cSum.x += c.x * A;
+ cSum.y += c.y * A;
+ cSum.z += c.z * A;
+ cSum.w += c.w * A;
i++;
}
Until(i >= a)
- c.r = cSum.r;
- c.g = cSum.g;
- c.b = cSum.b;
- c.a = cSum.a;
+ c.x = cSum.x;
+ c.y = cSum.y;
+ c.z = cSum.z;
+ c.w = cSum.w;
}
}
- void SamplerCore::sampleFloat(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD)
+ void SamplerCore::sampleFloat(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD)
{
if(state.textureType != TEXTURE_3D)
{
@@ -1035,7 +1035,7 @@
}
}
- void SamplerCore::sampleFloat2D(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &z, Float &lod, Int face[4], bool secondLOD)
+ void SamplerCore::sampleFloat2D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &z, Float &lod, Int face[4], bool secondLOD)
{
int componentCount = textureComponentCount();
bool gather = state.textureFilter == FILTER_GATHER;
@@ -1057,10 +1057,10 @@
}
else
{
- Color4f c0;
- Color4f c1;
- Color4f c2;
- Color4f c3;
+ Vector4f c0;
+ Vector4f c1;
+ Vector4f c2;
+ Vector4f c3;
Short4 uuuu0 = offsetSample(uuuu, mipmap, OFFSET(Mipmap,uHalf), (AddressingMode)state.addressingModeU == ADDRESSING_WRAP, gather ? 0 : -1);
Short4 vvvv0 = offsetSample(vvvv, mipmap, OFFSET(Mipmap,vHalf), (AddressingMode)state.addressingModeV == ADDRESSING_WRAP, gather ? 0 : -1);
@@ -1075,35 +1075,35 @@
if(!gather) // Blend
{
// Fractions
- Float4 fu = Fraction(Float4(As<UShort4>(uuuu0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fWidth)));
- Float4 fv = Fraction(Float4(As<UShort4>(vvvv0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fHeight)));
+ Float4 fu = Frac(Float4(As<UShort4>(uuuu0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fWidth)));
+ Float4 fv = Frac(Float4(As<UShort4>(vvvv0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fHeight)));
- if(componentCount >= 1) c0.r = c0.r + fu * (c1.r - c0.r);
- if(componentCount >= 2) c0.g = c0.g + fu * (c1.g - c0.g);
- if(componentCount >= 3) c0.b = c0.b + fu * (c1.b - c0.b);
- if(componentCount >= 4) c0.a = c0.a + fu * (c1.a - c0.a);
+ if(componentCount >= 1) c0.x = c0.x + fu * (c1.x - c0.x);
+ if(componentCount >= 2) c0.y = c0.y + fu * (c1.y - c0.y);
+ if(componentCount >= 3) c0.z = c0.z + fu * (c1.z - c0.z);
+ if(componentCount >= 4) c0.w = c0.w + fu * (c1.w - c0.w);
- if(componentCount >= 1) c2.r = c2.r + fu * (c3.r - c2.r);
- if(componentCount >= 2) c2.g = c2.g + fu * (c3.g - c2.g);
- if(componentCount >= 3) c2.b = c2.b + fu * (c3.b - c2.b);
- if(componentCount >= 4) c2.a = c2.a + fu * (c3.a - c2.a);
+ if(componentCount >= 1) c2.x = c2.x + fu * (c3.x - c2.x);
+ if(componentCount >= 2) c2.y = c2.y + fu * (c3.y - c2.y);
+ if(componentCount >= 3) c2.z = c2.z + fu * (c3.z - c2.z);
+ if(componentCount >= 4) c2.w = c2.w + fu * (c3.w - c2.w);
- if(componentCount >= 1) c.r = c0.r + fv * (c2.r - c0.r);
- if(componentCount >= 2) c.g = c0.g + fv * (c2.g - c0.g);
- if(componentCount >= 3) c.b = c0.b + fv * (c2.b - c0.b);
- if(componentCount >= 4) c.a = c0.a + fv * (c2.a - c0.a);
+ if(componentCount >= 1) c.x = c0.x + fv * (c2.x - c0.x);
+ if(componentCount >= 2) c.y = c0.y + fv * (c2.y - c0.y);
+ if(componentCount >= 3) c.z = c0.z + fv * (c2.z - c0.z);
+ if(componentCount >= 4) c.w = c0.w + fv * (c2.w - c0.w);
}
else
{
- c.r = c1.r;
- c.g = c2.r;
- c.b = c3.r;
- c.a = c0.r;
+ c.x = c1.x;
+ c.y = c2.x;
+ c.z = c3.x;
+ c.w = c0.x;
}
}
}
- void SamplerCore::sampleFloat3D(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD)
+ void SamplerCore::sampleFloat3D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD)
{
int componentCount = textureComponentCount();
@@ -1127,14 +1127,14 @@
}
else
{
- Color4f &c0 = c;
- Color4f c1;
- Color4f c2;
- Color4f c3;
- Color4f c4;
- Color4f c5;
- Color4f c6;
- Color4f c7;
+ Vector4f &c0 = c;
+ Vector4f c1;
+ Vector4f c2;
+ Vector4f c3;
+ Vector4f c4;
+ Vector4f c5;
+ Vector4f c6;
+ Vector4f c7;
Short4 uuuu0 = offsetSample(uuuu, mipmap, OFFSET(Mipmap,uHalf), (AddressingMode)state.addressingModeU == ADDRESSING_WRAP, -1);
Short4 vvvv0 = offsetSample(vvvv, mipmap, OFFSET(Mipmap,vHalf), (AddressingMode)state.addressingModeV == ADDRESSING_WRAP, -1);
@@ -1153,51 +1153,51 @@
sampleTexel(c7, uuuu1, vvvv1, wwww1, w, mipmap, buffer);
// Fractions
- Float4 fu = Fraction(Float4(As<UShort4>(uuuu0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fWidth)));
- Float4 fv = Fraction(Float4(As<UShort4>(vvvv0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fHeight)));
- Float4 fw = Fraction(Float4(As<UShort4>(wwww0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fDepth)));
+ Float4 fu = Frac(Float4(As<UShort4>(uuuu0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fWidth)));
+ Float4 fv = Frac(Float4(As<UShort4>(vvvv0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fHeight)));
+ Float4 fw = Frac(Float4(As<UShort4>(wwww0)) * *Pointer<Float4>(mipmap + OFFSET(Mipmap,fDepth)));
// Blend first slice
- if(componentCount >= 1) c0.r = c0.r + fu * (c1.r - c0.r);
- if(componentCount >= 2) c0.g = c0.g + fu * (c1.g - c0.g);
- if(componentCount >= 3) c0.b = c0.b + fu * (c1.b - c0.b);
- if(componentCount >= 4) c0.a = c0.a + fu * (c1.a - c0.a);
+ if(componentCount >= 1) c0.x = c0.x + fu * (c1.x - c0.x);
+ if(componentCount >= 2) c0.y = c0.y + fu * (c1.y - c0.y);
+ if(componentCount >= 3) c0.z = c0.z + fu * (c1.z - c0.z);
+ if(componentCount >= 4) c0.w = c0.w + fu * (c1.w - c0.w);
- if(componentCount >= 1) c2.r = c2.r + fu * (c3.r - c2.r);
- if(componentCount >= 2) c2.g = c2.g + fu * (c3.g - c2.g);
- if(componentCount >= 3) c2.b = c2.b + fu * (c3.b - c2.b);
- if(componentCount >= 4) c2.a = c2.a + fu * (c3.a - c2.a);
+ if(componentCount >= 1) c2.x = c2.x + fu * (c3.x - c2.x);
+ if(componentCount >= 2) c2.y = c2.y + fu * (c3.y - c2.y);
+ if(componentCount >= 3) c2.z = c2.z + fu * (c3.z - c2.z);
+ if(componentCount >= 4) c2.w = c2.w + fu * (c3.w - c2.w);
- if(componentCount >= 1) c0.r = c0.r + fv * (c2.r - c0.r);
- if(componentCount >= 2) c0.g = c0.g + fv * (c2.g - c0.g);
- if(componentCount >= 3) c0.b = c0.b + fv * (c2.b - c0.b);
- if(componentCount >= 4) c0.a = c0.a + fv * (c2.a - c0.a);
+ if(componentCount >= 1) c0.x = c0.x + fv * (c2.x - c0.x);
+ if(componentCount >= 2) c0.y = c0.y + fv * (c2.y - c0.y);
+ if(componentCount >= 3) c0.z = c0.z + fv * (c2.z - c0.z);
+ if(componentCount >= 4) c0.w = c0.w + fv * (c2.w - c0.w);
// Blend second slice
- if(componentCount >= 1) c4.r = c4.r + fu * (c5.r - c4.r);
- if(componentCount >= 2) c4.g = c4.g + fu * (c5.g - c4.g);
- if(componentCount >= 3) c4.b = c4.b + fu * (c5.b - c4.b);
- if(componentCount >= 4) c4.a = c4.a + fu * (c5.a - c4.a);
+ if(componentCount >= 1) c4.x = c4.x + fu * (c5.x - c4.x);
+ if(componentCount >= 2) c4.y = c4.y + fu * (c5.y - c4.y);
+ if(componentCount >= 3) c4.z = c4.z + fu * (c5.z - c4.z);
+ if(componentCount >= 4) c4.w = c4.w + fu * (c5.w - c4.w);
- if(componentCount >= 1) c6.r = c6.r + fu * (c7.r - c6.r);
- if(componentCount >= 2) c6.g = c6.g + fu * (c7.g - c6.g);
- if(componentCount >= 3) c6.b = c6.b + fu * (c7.b - c6.b);
- if(componentCount >= 4) c6.a = c6.a + fu * (c7.a - c6.a);
+ if(componentCount >= 1) c6.x = c6.x + fu * (c7.x - c6.x);
+ if(componentCount >= 2) c6.y = c6.y + fu * (c7.y - c6.y);
+ if(componentCount >= 3) c6.z = c6.z + fu * (c7.z - c6.z);
+ if(componentCount >= 4) c6.w = c6.w + fu * (c7.w - c6.w);
- if(componentCount >= 1) c4.r = c4.r + fv * (c6.r - c4.r);
- if(componentCount >= 2) c4.g = c4.g + fv * (c6.g - c4.g);
- if(componentCount >= 3) c4.b = c4.b + fv * (c6.b - c4.b);
- if(componentCount >= 4) c4.a = c4.a + fv * (c6.a - c4.a);
+ if(componentCount >= 1) c4.x = c4.x + fv * (c6.x - c4.x);
+ if(componentCount >= 2) c4.y = c4.y + fv * (c6.y - c4.y);
+ if(componentCount >= 3) c4.z = c4.z + fv * (c6.z - c4.z);
+ if(componentCount >= 4) c4.w = c4.w + fv * (c6.w - c4.w);
// Blend slices
- if(componentCount >= 1) c0.r = c0.r + fw * (c4.r - c0.r);
- if(componentCount >= 2) c0.g = c0.g + fw * (c4.g - c0.g);
- if(componentCount >= 3) c0.b = c0.b + fw * (c4.b - c0.b);
- if(componentCount >= 4) c0.a = c0.a + fw * (c4.a - c0.a);
+ if(componentCount >= 1) c0.x = c0.x + fw * (c4.x - c0.x);
+ if(componentCount >= 2) c0.y = c0.y + fw * (c4.y - c0.y);
+ if(componentCount >= 3) c0.z = c0.z + fw * (c4.z - c0.z);
+ if(componentCount >= 4) c0.w = c0.w + fw * (c4.w - c0.w);
}
}
- void SamplerCore::computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &uuuu, Float4 &vvvv, Float &lodBias, Color4f &dsx, Color4f &dsy, bool bias, bool gradients, bool lodProvided)
+ void SamplerCore::computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &uuuu, Float4 &vvvv, Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided)
{
if(!lodProvided)
{
@@ -1267,7 +1267,7 @@
lod = Min(lod, Float(MIPMAP_LEVELS - 2)); // Trilinear accesses lod+1
}
- void SamplerCore::computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &uuuu, Float4 &vvvv, Float4 &wwww, Float &lodBias, Color4f &dsx, Color4f &dsy, bool bias, bool gradients, bool lodProvided)
+ void SamplerCore::computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &uuuu, Float4 &vvvv, Float4 &wwww, Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided)
{
if(state.mipmapFilter == MIPMAP_NONE)
{
@@ -1337,9 +1337,9 @@
void SamplerCore::cubeFace(Int face[4], Float4 &U, Float4 &V, Float4 &lodU, Float4 &lodV, Float4 &x, Float4 &y, Float4 &z)
{
- Int4 xp = CmpNLE(x, Float4(0.0f, 0.0f, 0.0f, 0.0f)); // x > 0
- Int4 yp = CmpNLE(y, Float4(0.0f, 0.0f, 0.0f, 0.0f)); // y > 0
- Int4 zp = CmpNLE(z, Float4(0.0f, 0.0f, 0.0f, 0.0f)); // z > 0
+ Int4 xp = CmpNLE(x, Float4(0.0f)); // x > 0
+ Int4 yp = CmpNLE(y, Float4(0.0f)); // y > 0
+ Int4 zp = CmpNLE(z, Float4(0.0f)); // z > 0
Float4 absX = Abs(x);
Float4 absY = Abs(y);
@@ -1383,8 +1383,8 @@
Float4 M = As<Float4>((xyz & As<Int4>(x)) | (yzx & As<Int4>(y)) | (zxy & As<Int4>(z)));
M = reciprocal(M);
- U *= M * Float4(0.5f, 0.5f, 0.5f, 0.5f);
- V *= M * Float4(0.5f, 0.5f, 0.5f, 0.5f);
+ U *= M * Float4(0.5f);
+ V *= M * Float4(0.5f);
// Project coordinates onto one face for consistent LOD calculation
{
@@ -1404,8 +1404,8 @@
Float4 M = As<Float4>((xyz & As<Int4>(x)) | (yzx & As<Int4>(y)) | (zxy & As<Int4>(z)));
M = Rcp_pp(M);
- lodU *= M * Float4(0.5f, 0.5f, 0.5f, 0.5f);
- lodV *= M * Float4(0.5f, 0.5f, 0.5f, 0.5f);
+ lodU *= M * Float4(0.5f);
+ lodV *= M * Float4(0.5f);
}
}
@@ -1467,7 +1467,7 @@
index[3] = Extract(As<Int2>(uuu2), 1);
}
- void SamplerCore::sampleTexel(Color4i &c, Short4 &uuuu, Short4 &vvvv, Short4 &wwww, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4])
+ void SamplerCore::sampleTexel(Vector4i &c, Short4 &uuuu, Short4 &vvvv, Short4 &wwww, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4])
{
Int index[4];
@@ -1493,32 +1493,32 @@
Byte8 c1 = *Pointer<Byte8>(buffer[f1] + 4 * index[1]);
Byte8 c2 = *Pointer<Byte8>(buffer[f2] + 4 * index[2]);
Byte8 c3 = *Pointer<Byte8>(buffer[f3] + 4 * index[3]);
- c.r = UnpackLow(c0, c1);
- c.g = UnpackLow(c2, c3);
+ c.x = UnpackLow(c0, c1);
+ c.y = UnpackLow(c2, c3);
switch(state.textureFormat)
{
case FORMAT_A8R8G8B8:
- c.b = c.r;
- c.b = As<Short4>(UnpackLow(c.b, c.g));
- c.r = As<Short4>(UnpackHigh(c.r, c.g));
- c.g = c.b;
- c.a = c.r;
- c.b = UnpackLow(As<Byte8>(c.b), As<Byte8>(c.b));
- c.g = UnpackHigh(As<Byte8>(c.g), As<Byte8>(c.g));
- c.r = UnpackLow(As<Byte8>(c.r), As<Byte8>(c.r));
- c.a = UnpackHigh(As<Byte8>(c.a), As<Byte8>(c.a));
+ c.z = c.x;
+ c.z = As<Short4>(UnpackLow(c.z, c.y));
+ c.x = As<Short4>(UnpackHigh(c.x, c.y));
+ c.y = c.z;
+ c.w = c.x;
+ c.z = UnpackLow(As<Byte8>(c.z), As<Byte8>(c.z));
+ c.y = UnpackHigh(As<Byte8>(c.y), As<Byte8>(c.y));
+ c.x = UnpackLow(As<Byte8>(c.x), As<Byte8>(c.x));
+ c.w = UnpackHigh(As<Byte8>(c.w), As<Byte8>(c.w));
break;
case FORMAT_Q8W8V8U8:
- c.b = c.r;
- c.r = As<Short4>(UnpackLow(c.r, c.g));
- c.b = As<Short4>(UnpackHigh(c.b, c.g));
- c.g = c.r;
- c.a = c.b;
- c.r = UnpackLow(As<Byte8>(c.r), As<Byte8>(c.r));
- c.g = UnpackHigh(As<Byte8>(c.g), As<Byte8>(c.g));
- c.b = UnpackLow(As<Byte8>(c.b), As<Byte8>(c.b));
- c.a = UnpackHigh(As<Byte8>(c.a), As<Byte8>(c.a));
+ c.z = c.x;
+ c.x = As<Short4>(UnpackLow(c.x, c.y));
+ c.z = As<Short4>(UnpackHigh(c.z, c.y));
+ c.y = c.x;
+ c.w = c.z;
+ c.x = UnpackLow(As<Byte8>(c.x), As<Byte8>(c.x));
+ c.y = UnpackHigh(As<Byte8>(c.y), As<Byte8>(c.y));
+ c.z = UnpackLow(As<Byte8>(c.z), As<Byte8>(c.z));
+ c.w = UnpackHigh(As<Byte8>(c.w), As<Byte8>(c.w));
break;
default:
ASSERT(false);
@@ -1531,30 +1531,30 @@
Byte8 c1 = *Pointer<Byte8>(buffer[f1] + 4 * index[1]);
Byte8 c2 = *Pointer<Byte8>(buffer[f2] + 4 * index[2]);
Byte8 c3 = *Pointer<Byte8>(buffer[f3] + 4 * index[3]);
- c.r = UnpackLow(c0, c1);
- c.g = UnpackLow(c2, c3);
+ c.x = UnpackLow(c0, c1);
+ c.y = UnpackLow(c2, c3);
switch(state.textureFormat)
{
case FORMAT_X8R8G8B8:
- c.b = c.r;
- c.b = As<Short4>(UnpackLow(c.b, c.g));
- c.r = As<Short4>(UnpackHigh(c.r, c.g));
- c.g = c.b;
- c.b = UnpackLow(As<Byte8>(c.b), As<Byte8>(c.b));
- c.g = UnpackHigh(As<Byte8>(c.g), As<Byte8>(c.g));
- c.r = UnpackLow(As<Byte8>(c.r), As<Byte8>(c.r));
+ c.z = c.x;
+ c.z = As<Short4>(UnpackLow(c.z, c.y));
+ c.x = As<Short4>(UnpackHigh(c.x, c.y));
+ c.y = c.z;
+ c.z = UnpackLow(As<Byte8>(c.z), As<Byte8>(c.z));
+ c.y = UnpackHigh(As<Byte8>(c.y), As<Byte8>(c.y));
+ c.x = UnpackLow(As<Byte8>(c.x), As<Byte8>(c.x));
break;
case FORMAT_X8L8V8U8:
- c.b = c.r;
- c.r = As<Short4>(UnpackLow(c.r, c.g));
- c.b = As<Short4>(UnpackHigh(c.b, c.g));
- c.g = c.r;
- c.r = UnpackLow(As<Byte8>(c.r), As<Byte8>(Short4(0x0000, 0x0000, 0x0000, 0x0000)));
- c.r = c.r << 8;
- c.g = UnpackHigh(As<Byte8>(c.g), As<Byte8>(Short4(0x0000, 0x0000, 0x0000, 0x0000)));
- c.g = c.g << 8;
- c.b = UnpackLow(As<Byte8>(c.b), As<Byte8>(c.b));
+ c.z = c.x;
+ c.x = As<Short4>(UnpackLow(c.x, c.y));
+ c.z = As<Short4>(UnpackHigh(c.z, c.y));
+ c.y = c.x;
+ c.x = UnpackLow(As<Byte8>(c.x), As<Byte8>(Short4(0x0000, 0x0000, 0x0000, 0x0000)));
+ c.x = c.x << 8;
+ c.y = UnpackHigh(As<Byte8>(c.y), As<Byte8>(Short4(0x0000, 0x0000, 0x0000, 0x0000)));
+ c.y = c.y << 8;
+ c.z = UnpackLow(As<Byte8>(c.z), As<Byte8>(c.z));
break;
default:
ASSERT(false);
@@ -1562,10 +1562,10 @@
}
break;
case 2:
- c.r = Insert(c.r, *Pointer<Short>(buffer[f0] + 2 * index[0]), 0);
- c.r = Insert(c.r, *Pointer<Short>(buffer[f1] + 2 * index[1]), 1);
- c.r = Insert(c.r, *Pointer<Short>(buffer[f2] + 2 * index[2]), 2);
- c.r = Insert(c.r, *Pointer<Short>(buffer[f3] + 2 * index[3]), 3);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f0] + 2 * index[0]), 0);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f1] + 2 * index[1]), 1);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f2] + 2 * index[2]), 2);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f3] + 2 * index[3]), 3);
switch(state.textureFormat)
{
@@ -1573,8 +1573,8 @@
case FORMAT_V8U8:
case FORMAT_A8L8:
// FIXME: Unpack properly to 0.16 format
- c.g = c.r;
- c.r = c.r << 8;
+ c.y = c.x;
+ c.x = c.x << 8;
break;
default:
ASSERT(false);
@@ -1586,8 +1586,7 @@
c2 = Int(*Pointer<Byte>(buffer[f2] + index[2]));
c3 = Int(*Pointer<Byte>(buffer[f3] + index[3]));
c0 = c0 | (c1 << 8) | (c2 << 16) | (c3 << 24);
- c.r = As<Short4>(Int2(c0));
- c.r = UnpackLow(As<Byte8>(c.r), As<Byte8>(c.r));
+ c.x = Unpack(As<Byte4>(c0));
break;
default:
ASSERT(false);
@@ -1598,26 +1597,26 @@
switch(textureComponentCount())
{
case 4:
- c.r = *Pointer<Short4>(buffer[f0] + 8 * index[0]);
- c.g = *Pointer<Short4>(buffer[f1] + 8 * index[1]);
- c.b = *Pointer<Short4>(buffer[f2] + 8 * index[2]);
- c.a = *Pointer<Short4>(buffer[f3] + 8 * index[3]);
- transpose4x4(c.r, c.g, c.b, c.a);
+ c.x = *Pointer<Short4>(buffer[f0] + 8 * index[0]);
+ c.y = *Pointer<Short4>(buffer[f1] + 8 * index[1]);
+ c.z = *Pointer<Short4>(buffer[f2] + 8 * index[2]);
+ c.w = *Pointer<Short4>(buffer[f3] + 8 * index[3]);
+ transpose4x4(c.x, c.y, c.z, c.w);
break;
case 2:
- c.r = *Pointer<Short4>(buffer[f0] + 4 * index[0]);
- c.r = As<Short4>(UnpackLow(c.r, *Pointer<Short4>(buffer[f1] + 4 * index[1])));
- c.b = *Pointer<Short4>(buffer[f2] + 4 * index[2]);
- c.b = As<Short4>(UnpackLow(c.b, *Pointer<Short4>(buffer[f3] + 4 * index[3])));
- c.g = c.r;
- c.r = As<Short4>(UnpackLow(As<Int2>(c.r), As<Int2>(c.b)));
- c.g = As<Short4>(UnpackHigh(As<Int2>(c.g), As<Int2>(c.b)));
+ c.x = *Pointer<Short4>(buffer[f0] + 4 * index[0]);
+ c.x = As<Short4>(UnpackLow(c.x, *Pointer<Short4>(buffer[f1] + 4 * index[1])));
+ c.z = *Pointer<Short4>(buffer[f2] + 4 * index[2]);
+ c.z = As<Short4>(UnpackLow(c.z, *Pointer<Short4>(buffer[f3] + 4 * index[3])));
+ c.y = c.x;
+ c.x = As<Short4>(UnpackLow(As<Int2>(c.x), As<Int2>(c.z)));
+ c.y = As<Short4>(UnpackHigh(As<Int2>(c.y), As<Int2>(c.z)));
break;
case 1:
- c.r = Insert(c.r, *Pointer<Short>(buffer[f0] + 2 * index[0]), 0);
- c.r = Insert(c.r, *Pointer<Short>(buffer[f1] + 2 * index[1]), 1);
- c.r = Insert(c.r, *Pointer<Short>(buffer[f2] + 2 * index[2]), 2);
- c.r = Insert(c.r, *Pointer<Short>(buffer[f3] + 2 * index[3]), 3);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f0] + 2 * index[0]), 0);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f1] + 2 * index[1]), 1);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f2] + 2 * index[2]), 2);
+ c.x = Insert(c.x, *Pointer<Short>(buffer[f3] + 2 * index[3]), 3);
break;
default:
ASSERT(false);
@@ -1625,7 +1624,7 @@
}
}
- void SamplerCore::sampleTexel(Color4f &c, Short4 &uuuu, Short4 &vvvv, Short4 &wwww, Float4 &z, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4])
+ void SamplerCore::sampleTexel(Vector4f &c, Short4 &uuuu, Short4 &vvvv, Short4 &wwww, Float4 &z, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4])
{
Int index[4];
@@ -1640,34 +1639,34 @@
switch(textureComponentCount())
{
case 4:
- c.r = *Pointer<Float4>(buffer[f0] + index[0] * 16, 16);
- c.g = *Pointer<Float4>(buffer[f1] + index[1] * 16, 16);
- c.b = *Pointer<Float4>(buffer[f2] + index[2] * 16, 16);
- c.a = *Pointer<Float4>(buffer[f3] + index[3] * 16, 16);
- transpose4x4(c.r, c.g, c.b, c.a);
+ c.x = *Pointer<Float4>(buffer[f0] + index[0] * 16, 16);
+ c.y = *Pointer<Float4>(buffer[f1] + index[1] * 16, 16);
+ c.z = *Pointer<Float4>(buffer[f2] + index[2] * 16, 16);
+ c.w = *Pointer<Float4>(buffer[f3] + index[3] * 16, 16);
+ transpose4x4(c.x, c.y, c.z, c.w);
break;
case 2:
// FIXME: Optimal shuffling?
- c.r.xy = *Pointer<Float4>(buffer[f0] + index[0] * 8);
- c.r.zw = *Pointer<Float4>(buffer[f1] + index[1] * 8 - 8);
- c.b.xy = *Pointer<Float4>(buffer[f2] + index[2] * 8);
- c.b.zw = *Pointer<Float4>(buffer[f3] + index[3] * 8 - 8);
- c.g = c.r;
- c.r = Float4(c.r.xz, c.b.xz);
- c.g = Float4(c.g.yw, c.b.yw);
+ c.x.xy = *Pointer<Float4>(buffer[f0] + index[0] * 8);
+ c.x.zw = *Pointer<Float4>(buffer[f1] + index[1] * 8 - 8);
+ c.z.xy = *Pointer<Float4>(buffer[f2] + index[2] * 8);
+ c.z.zw = *Pointer<Float4>(buffer[f3] + index[3] * 8 - 8);
+ c.y = c.x;
+ c.x = Float4(c.x.xz, c.z.xz);
+ c.y = Float4(c.y.yw, c.z.yw);
break;
case 1:
// FIXME: Optimal shuffling?
- c.r.x = *Pointer<Float>(buffer[f0] + index[0] * 4);
- c.r.y = *Pointer<Float>(buffer[f1] + index[1] * 4);
- c.r.z = *Pointer<Float>(buffer[f2] + index[2] * 4);
- c.r.w = *Pointer<Float>(buffer[f3] + index[3] * 4);
+ c.x.x = *Pointer<Float>(buffer[f0] + index[0] * 4);
+ c.x.y = *Pointer<Float>(buffer[f1] + index[1] * 4);
+ c.x.z = *Pointer<Float>(buffer[f2] + index[2] * 4);
+ c.x.w = *Pointer<Float>(buffer[f3] + index[3] * 4);
if(state.textureFormat == FORMAT_D32F_SHADOW && state.textureFilter != FILTER_GATHER)
{
Float4 d = Min(Max(z, Float4(0.0f)), Float4(1.0f));
- c.r = As<Float4>(As<Int4>(CmpNLT(c.r, d)) & As<Int4>(Float4(1.0f, 1.0f, 1.0f, 1.0f))); // FIXME: Only less-equal?
+ c.x = As<Float4>(As<Int4>(CmpNLT(c.x, d)) & As<Int4>(Float4(1.0f))); // FIXME: Only less-equal?
}
break;
default:
@@ -1746,15 +1745,15 @@
void SamplerCore::convertFixed12(Short4 &ci, Float4 &cf)
{
- ci = RoundShort4(cf * Float4(0x1000, 0x1000, 0x1000, 0x1000));
+ ci = RoundShort4(cf * Float4(0x1000));
}
- void SamplerCore::convertFixed12(Color4i &ci, Color4f &cf)
+ void SamplerCore::convertFixed12(Vector4i &ci, Vector4f &cf)
{
- convertFixed12(ci.r, cf.r);
- convertFixed12(ci.g, cf.g);
- convertFixed12(ci.b, cf.b);
- convertFixed12(ci.a, cf.a);
+ convertFixed12(ci.x, cf.x);
+ convertFixed12(ci.y, cf.y);
+ convertFixed12(ci.z, cf.z);
+ convertFixed12(ci.w, cf.w);
}
void SamplerCore::convertSigned12(Float4 &cf, Short4 &ci)
@@ -1762,22 +1761,22 @@
cf = Float4(ci) * Float4(1.0f / 0x0FFE);
}
-// void SamplerCore::convertSigned12(Color4f &cf, Color4i &ci)
+// void SamplerCore::convertSigned12(Vector4f &cf, Vector4i &ci)
// {
-// convertSigned12(cf.r, ci.r);
-// convertSigned12(cf.g, ci.g);
-// convertSigned12(cf.b, ci.b);
-// convertSigned12(cf.a, ci.a);
+// convertSigned12(cf.x, ci.x);
+// convertSigned12(cf.y, ci.y);
+// convertSigned12(cf.z, ci.z);
+// convertSigned12(cf.w, ci.w);
// }
void SamplerCore::convertSigned15(Float4 &cf, Short4 &ci)
{
- cf = Float4(ci) * Float4(1.0f / 0x7FFF, 1.0f / 0x7FFF, 1.0f / 0x7FFF, 1.0f / 0x7FFF);
+ cf = Float4(ci) * Float4(1.0f / 0x7FFF);
}
void SamplerCore::convertUnsigned16(Float4 &cf, Short4 &ci)
{
- cf = Float4(As<UShort4>(ci)) * Float4(1.0f / 0xFFFF, 1.0f / 0xFFFF, 1.0f / 0xFFFF, 1.0f / 0xFFFF);
+ cf = Float4(As<UShort4>(ci)) * Float4(1.0f / 0xFFFF);
}
void SamplerCore::sRGBtoLinear16_12(Short4 &c)
diff --git a/src/Shader/SamplerCore.hpp b/src/Shader/SamplerCore.hpp
index c3c11e9..c03e30c 100644
--- a/src/Shader/SamplerCore.hpp
+++ b/src/Shader/SamplerCore.hpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -22,34 +22,34 @@
public:
SamplerCore(Pointer<Byte> &r, const Sampler::State &state);
- void sampleTexture(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool bias = false, bool fixed12 = true, bool gradients = false, bool lodProvided = false);
- void sampleTexture(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Color4f &dsx, Color4f &dsy, bool bias = false, bool gradients = false, bool lodProvided = false);
+ void sampleTexture(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias = false, bool fixed12 = true, bool gradients = false, bool lodProvided = false);
+ void sampleTexture(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float4 &q, Vector4f &dsx, Vector4f &dsy, bool bias = false, bool gradients = false, bool lodProvided = false);
private:
void border(Short4 &mask, Float4 &coordinates);
void border(Int4 &mask, Float4 &coordinates);
Short4 offsetSample(Short4 &uvw, Pointer<Byte> &mipmap, int halfOffset, bool wrap, int count);
- void sampleFilter(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided);
- void sampleAniso(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided);
- void sampleQuad(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
- void sampleQuad2D(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float &lod, Int face[4], bool secondLOD);
- void sample3D(Pointer<Byte> &texture, Color4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD);
- void sampleFloatFilter(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided);
- void sampleFloatAniso(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided);
- void sampleFloat(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
- void sampleFloat2D(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &z, Float &lod, Int face[4], bool secondLOD);
- void sampleFloat3D(Pointer<Byte> &texture, Color4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD);
- void computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &u, Float4 &v, Float &lodBias, Color4f &dsx, Color4f &dsy, bool bias, bool gradients, bool lodProvided);
- void computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &u, Float4 &v, Float4 &w, Float &lodBias, Color4f &dsx, Color4f &dsy, bool bias, bool gradients, bool lodProvided);
+ void sampleFilter(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided);
+ void sampleAniso(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided);
+ void sampleQuad(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
+ void sampleQuad2D(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float &lod, Int face[4], bool secondLOD);
+ void sample3D(Pointer<Byte> &texture, Vector4i &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD);
+ void sampleFloatFilter(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool lodProvided);
+ void sampleFloatAniso(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Int face[4], bool secondLOD, bool lodProvided);
+ void sampleFloat(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, Int face[4], bool secondLOD);
+ void sampleFloat2D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &z, Float &lod, Int face[4], bool secondLOD);
+ void sampleFloat3D(Pointer<Byte> &texture, Vector4f &c, Float4 &u, Float4 &v, Float4 &w, Float &lod, bool secondLOD);
+ void computeLod(Pointer<Byte> &texture, Float &lod, Float &anisotropy, Float4 &uDelta, Float4 &vDelta, Float4 &u, Float4 &v, Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided);
+ void computeLod3D(Pointer<Byte> &texture, Float &lod, Float4 &u, Float4 &v, Float4 &w, Float &lodBias, Vector4f &dsx, Vector4f &dsy, bool bias, bool gradients, bool lodProvided);
void cubeFace(Int face[4], Float4 &U, Float4 &V, Float4 &lodU, Float4 &lodV, Float4 &x, Float4 &y, Float4 &z);
void computeIndices(Int index[4], Short4 uuuu, Short4 vvvv, Short4 wwww, const Pointer<Byte> &mipmap);
- void sampleTexel(Color4i &c, Short4 &u, Short4 &v, Short4 &s, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4]);
- void sampleTexel(Color4f &c, Short4 &u, Short4 &v, Short4 &s, Float4 &z, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4]);
+ void sampleTexel(Vector4i &c, Short4 &u, Short4 &v, Short4 &s, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4]);
+ void sampleTexel(Vector4f &c, Short4 &u, Short4 &v, Short4 &s, Float4 &z, Pointer<Byte> &mipmap, Pointer<Byte> buffer[4]);
void selectMipmap(Pointer<Byte> &texture, Pointer<Byte> buffer[4], Pointer<Byte> &mipmap, Float &lod, Int face[4], bool secondLOD);
void address(Short4 &uuuu, Float4 &uw, AddressingMode addressingMode);
void convertFixed12(Short4 &ci, Float4 &cf);
- void convertFixed12(Color4i &ci, Color4f &cf);
+ void convertFixed12(Vector4i &ci, Vector4f &cf);
void convertSigned12(Float4 &cf, Short4 &ci);
void convertSigned15(Float4 &cf, Short4 &ci);
void convertUnsigned16(Float4 &cf, Short4 &ci);
diff --git a/src/Shader/SetupRoutine.cpp b/src/Shader/SetupRoutine.cpp
index 40a2a18..d77a21e 100644
--- a/src/Shader/SetupRoutine.cpp
+++ b/src/Shader/SetupRoutine.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -15,7 +15,7 @@
#include "Renderer/Primitive.hpp"
#include "Renderer/Polygon.hpp"
#include "Renderer/Renderer.hpp"
-#include "Reactor/Shell.hpp"
+#include "Reactor/Reactor.hpp"
namespace sw
{
@@ -108,10 +108,16 @@
if(state.twoSidedStencil)
{
- Byte8 clockwiseMask = IfThenElse(A > Float(0.0f), Byte8(0xFFFFFFFFFFFFFFFF), Byte8(0x0000000000000000));
-
- *Pointer<Byte8>(primitive + OFFSET(Primitive,clockwiseMask)) = clockwiseMask;
- *Pointer<Byte8>(primitive + OFFSET(Primitive,invClockwiseMask)) = ~clockwiseMask;
+ If(A > Float(0.0f))
+ {
+ *Pointer<Byte8>(primitive + OFFSET(Primitive,clockwiseMask)) = Byte8(0xFFFFFFFFFFFFFFFF);
+ *Pointer<Byte8>(primitive + OFFSET(Primitive,invClockwiseMask)) = Byte8(0x0000000000000000);
+ }
+ Else
+ {
+ *Pointer<Byte8>(primitive + OFFSET(Primitive,clockwiseMask)) = Byte8(0x0000000000000000);
+ *Pointer<Byte8>(primitive + OFFSET(Primitive,invClockwiseMask)) = Byte8(0xFFFFFFFFFFFFFFFF);
+ }
}
if(state.vFace)
@@ -145,8 +151,8 @@
Float w = v.w;
Float rhw = IfThenElse(w != Float(0.0f), Float(1.0f) / w, Float(1.0f));
- X[i] = RoundInt(*Pointer<Float>(data + OFFSET(DrawData,LLLLx16)) + v.x * rhw * *Pointer<Float>(data + OFFSET(DrawData,WWWWx16)));
- Y[i] = RoundInt(*Pointer<Float>(data + OFFSET(DrawData,TTTTx16)) + v.y * rhw * *Pointer<Float>(data + OFFSET(DrawData,HHHHx16)));
+ X[i] = RoundInt(*Pointer<Float>(data + OFFSET(DrawData,X0x16)) + v.x * rhw * *Pointer<Float>(data + OFFSET(DrawData,Wx16)));
+ Y[i] = RoundInt(*Pointer<Float>(data + OFFSET(DrawData,Y0x16)) + v.y * rhw * *Pointer<Float>(data + OFFSET(DrawData,Hx16)));
i++;
}
@@ -161,8 +167,8 @@
Do
{
- yMin = IfThenElse(Y[i] < yMin, Int(Y[i]), yMin); // FIXME: Min(Y[i], yMin)
- yMax = IfThenElse(Y[i] > yMax, Int(Y[i]), yMax); // FIXME: Max(Y[i], yMax)
+ yMin = Min(Y[i], yMin);
+ yMax = Max(Y[i], yMax);
i++;
}
@@ -183,6 +189,9 @@
{
Return(false);
}
+
+ yMin = Max(yMin, *Pointer<Int>(data + OFFSET(DrawData,scissorY0)));
+ yMax = Min(yMax, *Pointer<Int>(data + OFFSET(DrawData,scissorY1)));
For(Int q = 0, q < state.multiSample, q++)
{
@@ -229,7 +238,7 @@
Do
{
- edge(primitive, Int(Xq[i + 1 - d]), Int(Yq[i + 1 - d]), Int(Xq[i + d]), Int(Yq[i + d]), q);
+ edge(primitive, data, Int(Xq[i + 1 - d]), Int(Yq[i + 1 - d]), Int(Xq[i + d]), Int(Yq[i + d]), q);
i++;
}
@@ -372,7 +381,7 @@
// M[2].w = 0;
}
- if(state.perspective)
+ if(state.interpolateW)
{
Float4 ABC = M[0] + M[1] + M[2];
@@ -385,7 +394,7 @@
*Pointer<Float4>(primitive + OFFSET(Primitive,w.C), 16) = C;
}
- if(state.interpolateDepth)
+ if(state.interpolateZ)
{
Float z0 = *Pointer<Float>(v0 + OFFSET(Vertex,Z));
Float z1 = *Pointer<Float>(v1 + OFFSET(Vertex,Z));
@@ -491,7 +500,7 @@
if(component == 3) i.y = Float(1.0f);
if(component == 0) i.z = Float(0.5f);
- if(component == 1) i.z = Float(0.0f);
+ if(component == 1) i.z = Float(1.0f);
if(component == 2) i.z = Float(0.0f);
if(component == 3) i.z = Float(1.0f);
@@ -518,17 +527,9 @@
i *= w012;
}
- Float4 A;
- Float4 B;
- Float4 C;
-
- A = i.xxxx;
- B = i.yyyy;
- C = i.zzzz;
-
- A *= m[0];
- B *= m[1];
- C *= m[2];
+ Float4 A = i.xxxx * m[0];
+ Float4 B = i.yyyy * m[1];
+ Float4 C = i.zzzz * m[2];
C = A + B + C;
@@ -550,10 +551,13 @@
}
}
- void SetupRoutine::edge(Pointer<Byte> &primitive, Int &X1, Int &Y1, Int &X2, Int &Y2, Int &q)
+ void SetupRoutine::edge(Pointer<Byte> &primitive, Pointer<Byte> &data, Int &X1, Int &Y1, Int &X2, Int &Y2, Int &q)
{
If(Y1 != Y2)
{
+ Int xMin = *Pointer<Int>(data + OFFSET(DrawData,scissorX0));
+ Int xMax = *Pointer<Int>(data + OFFSET(DrawData,scissorX1));
+
Bool swap = Y2 < Y1;
Pointer<Byte> leftEdge = primitive + q * sizeof(Primitive) + OFFSET(Primitive,outline->left);
@@ -567,10 +571,10 @@
Y1 = IfThenElse(swap, Y2, Y1);
Y2 = IfThenElse(swap, Y0, Y2);
- Int y1 = (Y1 + 0x0000000F) >> 4;
- Int y2 = (Y2 + 0x0000000F) >> 4;
+ Int y1 = Max((Y1 + 0x0000000F) >> 4, *Pointer<Int>(data + OFFSET(DrawData,scissorY0)));
+ Int y2 = Min((Y2 + 0x0000000F) >> 4, *Pointer<Int>(data + OFFSET(DrawData,scissorY1)));
- If(y1 != y2)
+ If(y1 < y2)
{
// Deltas
Int DX12 = X2 - X1;
@@ -579,7 +583,7 @@
Int FDX12 = DX12 << 4;
Int FDY12 = DY12 << 4;
- Int X = DX12 * (-Y1 & 0xF) + X1 * DY12;
+ Int X = DX12 * ((y1 << 4) - Y1) + X1 * DY12;
Int x = X / FDY12; // Edge
Int d = X % FDY12; // Error-term
Int ceil = -d >> 31; // Ceiling division: remainder <= 0
@@ -597,7 +601,7 @@
Do
{
- *Pointer<Short>(edge + y * sizeof(Primitive::Span)) = Short(x);
+ *Pointer<Short>(edge + y * sizeof(Primitive::Span)) = Short(Clamp(x, xMin, xMax));
x += Q;
d += R;
diff --git a/src/Shader/SetupRoutine.hpp b/src/Shader/SetupRoutine.hpp
index cf40281..dd280c7 100644
--- a/src/Shader/SetupRoutine.hpp
+++ b/src/Shader/SetupRoutine.hpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -31,7 +31,7 @@
private:
void setupGradient(Pointer<Byte> &primitive, Pointer<Byte> &triangle, Float4 &w012, Float4 (&m)[3], Pointer<Byte> &v0, Pointer<Byte> &v1, Pointer<Byte> &v2, int attribute, int planeEquation, bool flatShading, bool sprite, bool perspective, bool wrap, int component);
- void edge(Pointer<Byte> &primitive, Int &X1, Int &Y1, Int &X2, Int &Y2, Int &q);
+ void edge(Pointer<Byte> &primitive, Pointer<Byte> &data, Int &X1, Int &Y1, Int &X2, Int &Y2, Int &q);
void conditionalRotate1(Bool condition, Pointer<Byte> &v0, Pointer<Byte> &v1, Pointer<Byte> &v2);
void conditionalRotate2(Bool condition, Pointer<Byte> &v0, Pointer<Byte> &v1, Pointer<Byte> &v2);
diff --git a/src/Shader/Shader.cpp b/src/Shader/Shader.cpp
index 43e8bdc..fcbc839 100644
--- a/src/Shader/Shader.cpp
+++ b/src/Shader/Shader.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -11,42 +11,137 @@
#include "Shader.hpp"
+#include "VertexShader.hpp"
+#include "PixelShader.hpp"
#include "Math.hpp"
#include "Debug.hpp"
-#include <stdarg.h>
#include <fstream>
#include <sstream>
namespace sw
{
- Shader::Instruction::Instruction()
+ volatile int Shader::serialCounter = 1;
+
+ Shader::Opcode Shader::OPCODE_DP(int i)
{
- operation.opcode = Operation::OPCODE_NOP;
- destinationParameter.type = Parameter::PARAMETER_VOID;
- sourceParameter[0].type = Parameter::PARAMETER_VOID;
- sourceParameter[1].type = Parameter::PARAMETER_VOID;
- sourceParameter[2].type = Parameter::PARAMETER_VOID;
- sourceParameter[3].type = Parameter::PARAMETER_VOID;
+ switch(i)
+ {
+ default: ASSERT(false);
+ case 1: return OPCODE_DP1;
+ case 2: return OPCODE_DP2;
+ case 3: return OPCODE_DP3;
+ case 4: return OPCODE_DP4;
+ }
}
- Shader::Instruction::Instruction(const unsigned long *token, int size, unsigned char majorVersion)
+ Shader::Opcode Shader::OPCODE_LEN(int i)
+ {
+ switch(i)
+ {
+ default: ASSERT(false);
+ case 1: return OPCODE_ABS;
+ case 2: return OPCODE_LEN2;
+ case 3: return OPCODE_LEN3;
+ case 4: return OPCODE_LEN4;
+ }
+ }
+
+ Shader::Opcode Shader::OPCODE_DIST(int i)
+ {
+ switch(i)
+ {
+ default: ASSERT(false);
+ case 1: return OPCODE_DIST1;
+ case 2: return OPCODE_DIST2;
+ case 3: return OPCODE_DIST3;
+ case 4: return OPCODE_DIST4;
+ }
+ }
+
+ Shader::Opcode Shader::OPCODE_NRM(int i)
+ {
+ switch(i)
+ {
+ default: ASSERT(false);
+ case 1: return OPCODE_SGN;
+ case 2: return OPCODE_NRM2;
+ case 3: return OPCODE_NRM3;
+ case 4: return OPCODE_NRM4;
+ }
+ }
+
+ Shader::Opcode Shader::OPCODE_FORWARD(int i)
+ {
+ switch(i)
+ {
+ default: ASSERT(false);
+ case 1: return OPCODE_FORWARD1;
+ case 2: return OPCODE_FORWARD2;
+ case 3: return OPCODE_FORWARD3;
+ case 4: return OPCODE_FORWARD4;
+ }
+ }
+
+ Shader::Opcode Shader::OPCODE_REFLECT(int i)
+ {
+ switch(i)
+ {
+ default: ASSERT(false);
+ case 1: return OPCODE_REFLECT1;
+ case 2: return OPCODE_REFLECT2;
+ case 3: return OPCODE_REFLECT3;
+ case 4: return OPCODE_REFLECT4;
+ }
+ }
+
+ Shader::Opcode Shader::OPCODE_REFRACT(int i)
+ {
+ switch(i)
+ {
+ default: ASSERT(false);
+ case 1: return OPCODE_REFRACT1;
+ case 2: return OPCODE_REFRACT2;
+ case 3: return OPCODE_REFRACT3;
+ case 4: return OPCODE_REFRACT4;
+ }
+ }
+
+ Shader::Instruction::Instruction(Opcode opcode) : opcode(opcode), analysis(0)
+ {
+ control = CONTROL_RESERVED0;
+
+ predicate = false;
+ predicateNot = false;
+ predicateSwizzle = 0xE4;
+
+ coissue = false;
+ samplerType = SAMPLER_UNKNOWN;
+ usage = USAGE_POSITION;
+ usageIndex = 0;
+ }
+
+ Shader::Instruction::Instruction(const unsigned long *token, int size, unsigned char majorVersion) : analysis(0)
{
parseOperationToken(*token++, majorVersion);
- if(operation.opcode == Operation::OPCODE_IF ||
- operation.opcode == Operation::OPCODE_IFC ||
- operation.opcode == Operation::OPCODE_LOOP ||
- operation.opcode == Operation::OPCODE_REP ||
- operation.opcode == Operation::OPCODE_BREAKC ||
- operation.opcode == Operation::OPCODE_BREAKP) // No destination operand
+ samplerType = SAMPLER_UNKNOWN;
+ usage = USAGE_POSITION;
+ usageIndex = 0;
+
+ if(opcode == OPCODE_IF ||
+ opcode == OPCODE_IFC ||
+ opcode == OPCODE_LOOP ||
+ opcode == OPCODE_REP ||
+ opcode == OPCODE_BREAKC ||
+ opcode == OPCODE_BREAKP) // No destination operand
{
if(size > 0) parseSourceToken(0, token++, majorVersion);
if(size > 1) parseSourceToken(1, token++, majorVersion);
if(size > 2) parseSourceToken(2, token++, majorVersion);
if(size > 3) ASSERT(false);
}
- else if(operation.opcode == Operation::OPCODE_DCL)
+ else if(opcode == OPCODE_DCL)
{
parseDeclarationToken(*token++);
parseDestinationToken(token++, majorVersion);
@@ -57,7 +152,7 @@
{
parseDestinationToken(token, majorVersion);
- if(destinationParameter.relative && majorVersion >= 3)
+ if(dst.rel.type != PARAMETER_VOID && majorVersion >= 3)
{
token++;
size--;
@@ -67,12 +162,12 @@
size--;
}
- if(operation.predicate)
+ if(predicate)
{
ASSERT(size != 0);
- operation.predicateNot = (SourceParameter::Modifier)((*token & 0x0F000000) >> 24) == SourceParameter::MODIFIER_NOT;
- operation.predicateSwizzle = (unsigned char)((*token & 0x00FF0000) >> 16);
+ predicateNot = (Modifier)((*token & 0x0F000000) >> 24) == MODIFIER_NOT;
+ predicateSwizzle = (unsigned char)((*token & 0x00FF0000) >> 16);
token++;
size--;
@@ -85,7 +180,7 @@
token++;
size--;
- if(sourceParameter[i].relative && majorVersion >= 2)
+ if(src[i].rel.type != PARAMETER_VOID && majorVersion >= 2)
{
token++;
size--;
@@ -98,105 +193,40 @@
{
}
- Shader::Instruction::Operation::Opcode Shader::Instruction::getOpcode() const
- {
- return operation.opcode;
- }
-
- const Shader::Instruction::DestinationParameter &Shader::Instruction::getDestinationParameter() const
- {
- return destinationParameter;
- }
-
- const Shader::Instruction::SourceParameter &Shader::Instruction::getSourceParameter(int i) const
- {
- return sourceParameter[i];
- }
-
- bool Shader::Instruction::isCoissue() const
- {
- return operation.coissue;
- }
-
- bool Shader::Instruction::isProject() const
- {
- return operation.project;
- }
-
- bool Shader::Instruction::isBias() const
- {
- return operation.bias;
- }
-
- bool Shader::Instruction::isPredicate() const
- {
- return operation.predicate;
- }
-
- bool Shader::Instruction::isPredicateNot() const
- {
- return operation.predicateNot;
- }
-
- unsigned char Shader::Instruction::getPredicateSwizzle() const
- {
- return operation.predicateSwizzle;
- }
-
- Shader::Instruction::Operation::Control Shader::Instruction::getControl() const
- {
- return operation.control;
- }
-
- Shader::Instruction::Operation::Usage Shader::Instruction::getUsage() const
- {
- return operation.usage;
- }
-
- unsigned char Shader::Instruction::getUsageIndex() const
- {
- return operation.usageIndex;
- }
-
- Shader::Instruction::Operation::SamplerType Shader::Instruction::getSamplerType() const
- {
- return operation.samplerType;
- }
-
std::string Shader::Instruction::string(ShaderType shaderType, unsigned short version) const
{
std::string instructionString;
- if(operation.opcode != Operation::OPCODE_DCL)
+ if(opcode != OPCODE_DCL)
{
- instructionString += operation.coissue ? "+ " : "";
+ instructionString += coissue ? "+ " : "";
- if(operation.predicate)
+ if(predicate)
{
- instructionString += operation.predicateNot ? "(!p0" : "(p0";
- instructionString += swizzleString(Parameter::PARAMETER_PREDICATE, operation.predicateSwizzle);
+ instructionString += predicateNot ? "(!p0" : "(p0";
+ instructionString += swizzleString(PARAMETER_PREDICATE, predicateSwizzle);
instructionString += ") ";
}
- instructionString += operation.string(version) + operation.controlString() + destinationParameter.shiftString() + destinationParameter.modifierString();
+ instructionString += operationString(version) + controlString() + dst.shiftString() + dst.modifierString();
- if(destinationParameter.type != Parameter::PARAMETER_VOID)
+ if(dst.type != PARAMETER_VOID)
{
- instructionString += " " + destinationParameter.string(shaderType, version) +
- destinationParameter.relativeString() +
- destinationParameter.maskString();
+ instructionString += " " + dst.string(shaderType, version) +
+ dst.relativeString() +
+ dst.maskString();
}
for(int i = 0; i < 4; i++)
{
- if(sourceParameter[i].type != Parameter::PARAMETER_VOID)
+ if(src[i].type != PARAMETER_VOID)
{
- instructionString += (destinationParameter.type != Parameter::PARAMETER_VOID || i > 0) ? ", " : " ";
- instructionString += sourceParameter[i].preModifierString() +
- sourceParameter[i].string(shaderType, version) +
- sourceParameter[i].relativeString() +
- sourceParameter[i].postModifierString() +
- sourceParameter[i].swizzleString();
+ instructionString += (dst.type != PARAMETER_VOID || i > 0) ? ", " : " ";
+ instructionString += src[i].preModifierString() +
+ src[i].string(shaderType, version) +
+ src[i].relativeString() +
+ src[i].postModifierString() +
+ src[i].swizzleString();
}
}
}
@@ -204,68 +234,68 @@
{
instructionString += "dcl";
- if(destinationParameter.type == Parameter::PARAMETER_SAMPLER)
+ if(dst.type == PARAMETER_SAMPLER)
{
- switch(operation.samplerType)
+ switch(samplerType)
{
- case Operation::SAMPLER_UNKNOWN: instructionString += " "; break;
- case Operation::SAMPLER_1D: instructionString += "_1d "; break;
- case Operation::SAMPLER_2D: instructionString += "_2d "; break;
- case Operation::SAMPLER_CUBE: instructionString += "_cube "; break;
- case Operation::SAMPLER_VOLUME: instructionString += "_volume "; break;
+ case SAMPLER_UNKNOWN: instructionString += " "; break;
+ case SAMPLER_1D: instructionString += "_1d "; break;
+ case SAMPLER_2D: instructionString += "_2d "; break;
+ case SAMPLER_CUBE: instructionString += "_cube "; break;
+ case SAMPLER_VOLUME: instructionString += "_volume "; break;
default:
ASSERT(false);
}
- instructionString += destinationParameter.string(shaderType, version);
+ instructionString += dst.string(shaderType, version);
}
- else if(destinationParameter.type == Parameter::PARAMETER_INPUT ||
- destinationParameter.type == Parameter::PARAMETER_OUTPUT ||
- destinationParameter.type == Parameter::PARAMETER_TEXTURE)
+ else if(dst.type == PARAMETER_INPUT ||
+ dst.type == PARAMETER_OUTPUT ||
+ dst.type == PARAMETER_TEXTURE)
{
if(version >= 0x0300)
{
- switch(operation.usage)
+ switch(usage)
{
- case Operation::USAGE_POSITION: instructionString += "_position"; break;
- case Operation::USAGE_BLENDWEIGHT: instructionString += "_blendweight"; break;
- case Operation::USAGE_BLENDINDICES: instructionString += "_blendindices"; break;
- case Operation::USAGE_NORMAL: instructionString += "_normal"; break;
- case Operation::USAGE_PSIZE: instructionString += "_psize"; break;
- case Operation::USAGE_TEXCOORD: instructionString += "_texcoord"; break;
- case Operation::USAGE_TANGENT: instructionString += "_tangent"; break;
- case Operation::USAGE_BINORMAL: instructionString += "_binormal"; break;
- case Operation::USAGE_TESSFACTOR: instructionString += "_tessfactor"; break;
- case Operation::USAGE_POSITIONT: instructionString += "_positiont"; break;
- case Operation::USAGE_COLOR: instructionString += "_color"; break;
- case Operation::USAGE_FOG: instructionString += "_fog"; break;
- case Operation::USAGE_DEPTH: instructionString += "_depth"; break;
- case Operation::USAGE_SAMPLE: instructionString += "_sample"; break;
+ case USAGE_POSITION: instructionString += "_position"; break;
+ case USAGE_BLENDWEIGHT: instructionString += "_blendweight"; break;
+ case USAGE_BLENDINDICES: instructionString += "_blendindices"; break;
+ case USAGE_NORMAL: instructionString += "_normal"; break;
+ case USAGE_PSIZE: instructionString += "_psize"; break;
+ case USAGE_TEXCOORD: instructionString += "_texcoord"; break;
+ case USAGE_TANGENT: instructionString += "_tangent"; break;
+ case USAGE_BINORMAL: instructionString += "_binormal"; break;
+ case USAGE_TESSFACTOR: instructionString += "_tessfactor"; break;
+ case USAGE_POSITIONT: instructionString += "_positiont"; break;
+ case USAGE_COLOR: instructionString += "_color"; break;
+ case USAGE_FOG: instructionString += "_fog"; break;
+ case USAGE_DEPTH: instructionString += "_depth"; break;
+ case USAGE_SAMPLE: instructionString += "_sample"; break;
default:
ASSERT(false);
}
- if(operation.usageIndex > 0)
+ if(usageIndex > 0)
{
std::ostringstream buffer;
- buffer << (int)operation.usageIndex;
+ buffer << (int)usageIndex;
instructionString += buffer.str();
}
}
- else ASSERT(destinationParameter.type != Parameter::PARAMETER_OUTPUT);
+ else ASSERT(dst.type != PARAMETER_OUTPUT);
instructionString += " ";
- instructionString += destinationParameter.string(shaderType, version);
- instructionString += destinationParameter.maskString();
+ instructionString += dst.string(shaderType, version);
+ instructionString += dst.maskString();
}
- else if(destinationParameter.type == Parameter::PARAMETER_MISCTYPE) // vPos and vFace
+ else if(dst.type == PARAMETER_MISCTYPE) // vPos and vFace
{
instructionString += " ";
- instructionString += destinationParameter.string(shaderType, version);
+ instructionString += dst.string(shaderType, version);
}
else ASSERT(false);
}
@@ -273,145 +303,7 @@
return instructionString;
}
- std::string Shader::Instruction::Operation::string(unsigned short version) const
- {
- switch(opcode)
- {
- case OPCODE_NOP: return "nop";
- case OPCODE_MOV: return "mov";
- case OPCODE_ADD: return "add";
- case OPCODE_SUB: return "sub";
- case OPCODE_MAD: return "mad";
- case OPCODE_MUL: return "mul";
- case OPCODE_RCP: return "rcp";
- case OPCODE_RSQ: return "rsq";
- case OPCODE_DP3: return "dp3";
- case OPCODE_DP4: return "dp4";
- case OPCODE_MIN: return "min";
- case OPCODE_MAX: return "max";
- case OPCODE_SLT: return "slt";
- case OPCODE_SGE: return "sge";
- case OPCODE_EXP: return "exp";
- case OPCODE_LOG: return "log";
- case OPCODE_LIT: return "lit";
- case OPCODE_DST: return "dst";
- case OPCODE_LRP: return "lrp";
- case OPCODE_FRC: return "frc";
- case OPCODE_M4X4: return "m4x4";
- case OPCODE_M4X3: return "m4x3";
- case OPCODE_M3X4: return "m3x4";
- case OPCODE_M3X3: return "m3x3";
- case OPCODE_M3X2: return "m3x2";
- case OPCODE_CALL: return "call";
- case OPCODE_CALLNZ: return "callnz";
- case OPCODE_LOOP: return "loop";
- case OPCODE_RET: return "ret";
- case OPCODE_ENDLOOP: return "endloop";
- case OPCODE_LABEL: return "label";
- case OPCODE_DCL: return "dcl";
- case OPCODE_POW: return "pow";
- case OPCODE_CRS: return "crs";
- case OPCODE_SGN: return "sgn";
- case OPCODE_ABS: return "abs";
- case OPCODE_NRM: return "nrm";
- case OPCODE_SINCOS: return "sincos";
- case OPCODE_REP: return "rep";
- case OPCODE_ENDREP: return "endrep";
- case OPCODE_IF: return "if";
- case OPCODE_IFC: return "ifc";
- case OPCODE_ELSE: return "else";
- case OPCODE_ENDIF: return "endif";
- case OPCODE_BREAK: return "break";
- case OPCODE_BREAKC: return "breakc";
- case OPCODE_MOVA: return "mova";
- case OPCODE_DEFB: return "defb";
- case OPCODE_DEFI: return "defi";
- case OPCODE_TEXCOORD: return "texcoord";
- case OPCODE_TEXKILL: return "texkill";
- case OPCODE_TEX:
- if(version < 0x0104) return "tex";
- else return "texld";
- case OPCODE_TEXBEM: return "texbem";
- case OPCODE_TEXBEML: return "texbeml";
- case OPCODE_TEXREG2AR: return "texreg2ar";
- case OPCODE_TEXREG2GB: return "texreg2gb";
- case OPCODE_TEXM3X2PAD: return "texm3x2pad";
- case OPCODE_TEXM3X2TEX: return "texm3x2tex";
- case OPCODE_TEXM3X3PAD: return "texm3x3pad";
- case OPCODE_TEXM3X3TEX: return "texm3x3tex";
- case OPCODE_RESERVED0: return "reserved0";
- case OPCODE_TEXM3X3SPEC: return "texm3x3spec";
- case OPCODE_TEXM3X3VSPEC: return "texm3x3vspec";
- case OPCODE_EXPP: return "expp";
- case OPCODE_LOGP: return "logp";
- case OPCODE_CND: return "cnd";
- case OPCODE_DEF: return "def";
- case OPCODE_TEXREG2RGB: return "texreg2rgb";
- case OPCODE_TEXDP3TEX: return "texdp3tex";
- case OPCODE_TEXM3X2DEPTH: return "texm3x2depth";
- case OPCODE_TEXDP3: return "texdp3";
- case OPCODE_TEXM3X3: return "texm3x3";
- case OPCODE_TEXDEPTH: return "texdepth";
- case OPCODE_CMP: return "cmp";
- case OPCODE_BEM: return "bem";
- case OPCODE_DP2ADD: return "dp2add";
- case OPCODE_DSX: return "dsx";
- case OPCODE_DSY: return "dsy";
- case OPCODE_TEXLDD: return "texldd";
- case OPCODE_SETP: return "setp";
- case OPCODE_TEXLDL: return "texldl";
- case OPCODE_BREAKP: return "breakp";
- case OPCODE_PHASE: return "phase";
- case OPCODE_COMMENT: return "comment";
- case OPCODE_END: return "end";
- case OPCODE_PS_1_0: return "ps_1_0";
- case OPCODE_PS_1_1: return "ps_1_1";
- case OPCODE_PS_1_2: return "ps_1_2";
- case OPCODE_PS_1_3: return "ps_1_3";
- case OPCODE_PS_1_4: return "ps_1_4";
- case OPCODE_PS_2_0: return "ps_2_0";
- case OPCODE_PS_2_x: return "ps_2_x";
- case OPCODE_PS_3_0: return "ps_3_0";
- case OPCODE_VS_1_0: return "vs_1_0";
- case OPCODE_VS_1_1: return "vs_1_1";
- case OPCODE_VS_2_0: return "vs_2_0";
- case OPCODE_VS_2_x: return "vs_2_x";
- case OPCODE_VS_2_sw: return "vs_2_sw";
- case OPCODE_VS_3_0: return "vs_3_0";
- case OPCODE_VS_3_sw: return "vs_3_sw";
- default:
- ASSERT(false);
- }
-
- return "<unknown>";
- }
-
- std::string Shader::Instruction::Operation::controlString() const
- {
- if(opcode != OPCODE_LOOP && opcode != OPCODE_BREAKC && opcode != OPCODE_IFC && opcode != OPCODE_SETP)
- {
- if(project) return "p";
-
- if(bias) return "b";
-
- // FIXME: LOD
- }
-
- switch(control)
- {
- case 1: return "_gt";
- case 2: return "_eq";
- case 3: return "_ge";
- case 4: return "_lt";
- case 5: return "_ne";
- case 6: return "_le";
- default:
- return "";
- // ASSERT(false); // FIXME
- }
- }
-
- std::string Shader::Instruction::DestinationParameter::modifierString() const
+ std::string Shader::DestinationParameter::modifierString() const
{
if(type == PARAMETER_VOID || type == PARAMETER_LABEL)
{
@@ -420,6 +312,11 @@
std::string modifierString;
+ if(integer)
+ {
+ modifierString += "_int";
+ }
+
if(saturate)
{
modifierString += "_sat";
@@ -438,7 +335,7 @@
return modifierString;
}
- std::string Shader::Instruction::DestinationParameter::shiftString() const
+ std::string Shader::DestinationParameter::shiftString() const
{
if(type == PARAMETER_VOID || type == PARAMETER_LABEL)
{
@@ -460,7 +357,7 @@
}
}
- std::string Shader::Instruction::DestinationParameter::maskString() const
+ std::string Shader::DestinationParameter::maskString() const
{
if(type == PARAMETER_VOID || type == PARAMETER_LABEL)
{
@@ -492,7 +389,7 @@
return "";
}
- std::string Shader::Instruction::SourceParameter::preModifierString() const
+ std::string Shader::SourceParameter::preModifierString() const
{
if(type == PARAMETER_VOID)
{
@@ -522,30 +419,48 @@
return "";
}
- std::string Shader::Instruction::Parameter::relativeString() const
+ std::string Shader::Parameter::relativeString() const
{
- if(!relative) return "";
-
- if(relativeType == Parameter::PARAMETER_ADDR)
+ if(type == PARAMETER_CONST || type == PARAMETER_INPUT || type == PARAMETER_OUTPUT || type == PARAMETER_TEMP)
{
- switch(relativeSwizzle & 0x03)
+ if(rel.type == PARAMETER_VOID)
{
- case 0: return "[a0.x]";
- case 1: return "[a0.y]";
- case 2: return "[a0.z]";
- case 3: return "[a0.w]";
+ return "";
}
+ else if(rel.type == PARAMETER_ADDR)
+ {
+ switch(rel.swizzle & 0x03)
+ {
+ case 0: return "[a0.x]";
+ case 1: return "[a0.y]";
+ case 2: return "[a0.z]";
+ case 3: return "[a0.w]";
+ }
+ }
+ else if(rel.type == PARAMETER_TEMP)
+ {
+ std::ostringstream buffer;
+ buffer << rel.index;
+
+ switch(rel.swizzle & 0x03)
+ {
+ case 0: return "[r" + buffer.str() + ".x]";
+ case 1: return "[r" + buffer.str() + ".y]";
+ case 2: return "[r" + buffer.str() + ".z]";
+ case 3: return "[r" + buffer.str() + ".w]";
+ }
+ }
+ else if(rel.type == PARAMETER_LOOP)
+ {
+ return "[aL]";
+ }
+ else ASSERT(false);
}
- else if(relativeType == Parameter::PARAMETER_LOOP)
- {
- return "[aL]";
- }
- else ASSERT(false);
return "";
}
- std::string Shader::Instruction::SourceParameter::postModifierString() const
+ std::string Shader::SourceParameter::postModifierString() const
{
if(type == PARAMETER_VOID)
{
@@ -575,7 +490,7 @@
return "";
}
- std::string Shader::Instruction::SourceParameter::swizzleString() const
+ std::string Shader::SourceParameter::swizzleString() const
{
return Instruction::swizzleString(type, swizzle);
}
@@ -584,19 +499,21 @@
{
if((token & 0xFFFF0000) == 0xFFFF0000 || (token & 0xFFFF0000) == 0xFFFE0000) // Version token
{
- operation.opcode = (Operation::Opcode)token;
- operation.predicate = false;
- operation.coissue = false;
+ opcode = (Opcode)token;
+
+ control = CONTROL_RESERVED0;
+ predicate = false;
+ coissue = false;
}
else
{
- operation.opcode = (Operation::Opcode)(token & 0x0000FFFF);
- operation.control = (Operation::Control)((token & 0x00FF0000) >> 16);
+ opcode = (Opcode)(token & 0x0000FFFF);
+ control = (Control)((token & 0x00FF0000) >> 16);
int size = (token & 0x0F000000) >> 24;
- operation.predicate = (token & 0x10000000) != 0x00000000;
- operation.coissue = (token & 0x40000000) != 0x00000000;
+ predicate = (token & 0x10000000) != 0x00000000;
+ coissue = (token & 0x40000000) != 0x00000000;
if(majorVersion < 2)
{
@@ -608,7 +525,7 @@
if(majorVersion < 2)
{
- if(operation.predicate)
+ if(predicate)
{
ASSERT(false);
}
@@ -621,7 +538,7 @@
if(majorVersion >= 2)
{
- if(operation.coissue)
+ if(coissue)
{
ASSERT(false); // Reserved
}
@@ -636,43 +553,44 @@
void Shader::Instruction::parseDeclarationToken(unsigned long token)
{
- operation.samplerType = (Operation::SamplerType)((token & 0x78000000) >> 27);
- operation.usage = (Operation::Usage)(token & 0x0000001F);
- operation.usageIndex = (unsigned char)((token & 0x000F0000) >> 16);
+ samplerType = (SamplerType)((token & 0x78000000) >> 27);
+ usage = (Usage)(token & 0x0000001F);
+ usageIndex = (unsigned char)((token & 0x000F0000) >> 16);
}
void Shader::Instruction::parseDestinationToken(const unsigned long *token, unsigned char majorVersion)
{
- destinationParameter.index = (unsigned short)(token[0] & 0x000007FF);
- destinationParameter.type = (Parameter::Type)(((token[0] & 0x00001800) >> 8) | ((token[0] & 0x70000000) >> 28));
+ dst.index = (unsigned short)(token[0] & 0x000007FF);
+ dst.type = (ParameterType)(((token[0] & 0x00001800) >> 8) | ((token[0] & 0x70000000) >> 28));
// TODO: Check type and index range
- destinationParameter.relative = (token[0] & 0x00002000) != 0x00000000;
- destinationParameter.relativeType = Parameter::PARAMETER_ADDR;
- destinationParameter.relativeSwizzle = 0x00;
+ bool relative = (token[0] & 0x00002000) != 0x00000000;
+ dst.rel.type = relative ? PARAMETER_ADDR : PARAMETER_VOID;
+ dst.rel.swizzle = 0x00;
+ dst.rel.scale = 1;
- if(destinationParameter.relative && majorVersion >= 3)
+ if(relative && majorVersion >= 3)
{
- destinationParameter.relativeType = (Parameter::Type)(((token[1] & 0x00001800) >> 8) | ((token[1] & 0x70000000) >> 28));
- destinationParameter.relativeSwizzle = (unsigned char)((token[1] & 0x00FF0000) >> 16);
+ dst.rel.type = (ParameterType)(((token[1] & 0x00001800) >> 8) | ((token[1] & 0x70000000) >> 28));
+ dst.rel.swizzle = (unsigned char)((token[1] & 0x00FF0000) >> 16);
}
- else if(destinationParameter.relative) ASSERT(false); // Reserved
+ else if(relative) ASSERT(false); // Reserved
if((token[0] & 0x0000C000) != 0x00000000)
{
ASSERT(false); // Reserved
}
- destinationParameter.mask = (unsigned char)((token[0] & 0x000F0000) >> 16);
- destinationParameter.saturate = (token[0] & 0x00100000) != 0;
- destinationParameter.partialPrecision = (token[0] & 0x00200000) != 0;
- destinationParameter.centroid = (token[0] & 0x00400000) != 0;
- destinationParameter.shift = (signed char)((token[0] & 0x0F000000) >> 20) >> 4;
+ dst.mask = (unsigned char)((token[0] & 0x000F0000) >> 16);
+ dst.saturate = (token[0] & 0x00100000) != 0;
+ dst.partialPrecision = (token[0] & 0x00200000) != 0;
+ dst.centroid = (token[0] & 0x00400000) != 0;
+ dst.shift = (signed char)((token[0] & 0x0F000000) >> 20) >> 4;
if(majorVersion >= 2)
{
- if(destinationParameter.shift)
+ if(dst.shift)
{
ASSERT(false); // Reserved
}
@@ -687,67 +605,71 @@
void Shader::Instruction::parseSourceToken(int i, const unsigned long *token, unsigned char majorVersion)
{
// Defaults
- sourceParameter[i].value = (float&)*token;
- sourceParameter[i].type = Parameter::PARAMETER_VOID;
- sourceParameter[i].modifier = SourceParameter::MODIFIER_NONE;
- sourceParameter[i].swizzle = 0xE4;
- sourceParameter[i].relative = false;
- sourceParameter[i].relativeType = Parameter::PARAMETER_ADDR;
- sourceParameter[i].relativeSwizzle = 0x00;
+ src[i].index = 0;
+ src[i].type = PARAMETER_VOID;
+ src[i].modifier = MODIFIER_NONE;
+ src[i].swizzle = 0xE4;
+ src[i].rel.type = PARAMETER_VOID;
+ src[i].rel.swizzle = 0x00;
+ src[i].rel.scale = 1;
- switch(operation.opcode)
+ switch(opcode)
{
- case Instruction::Operation::OPCODE_DEF:
- sourceParameter[i].type = Parameter::PARAMETER_FLOATLITERAL;
+ case OPCODE_DEF:
+ src[0].type = PARAMETER_FLOAT4LITERAL;
+ src[0].value[i] = *(float*)token;
break;
- case Instruction::Operation::OPCODE_DEFB:
- sourceParameter[i].type = Parameter::PARAMETER_BOOLLITERAL;
+ case OPCODE_DEFB:
+ src[0].type = PARAMETER_BOOL1LITERAL;
+ src[0].boolean[0] = *(int*)token;
break;
- case Instruction::Operation::OPCODE_DEFI:
- sourceParameter[i].type = Parameter::PARAMETER_INTLITERAL;
+ case OPCODE_DEFI:
+ src[0].type = PARAMETER_INT4LITERAL;
+ src[0].integer[i] = *(int*)token;
break;
default:
- sourceParameter[i].index = (unsigned short)(token[0] & 0x000007FF);
- sourceParameter[i].type = (Parameter::Type)(((token[0] & 0x00001800) >> 8) | ((token[0] & 0x70000000) >> 28));
+ src[i].index = (unsigned short)(token[0] & 0x000007FF);
+ src[i].type = (ParameterType)(((token[0] & 0x00001800) >> 8) | ((token[0] & 0x70000000) >> 28));
// FIXME: Check type and index range
- sourceParameter[i].relative = (token[0] & 0x00002000) != 0x00000000;
+ bool relative = (token[0] & 0x00002000) != 0x00000000;
+ src[i].rel.type = relative ? PARAMETER_ADDR : PARAMETER_VOID;
if((token[0] & 0x0000C000) != 0x00000000)
{
- if(operation.opcode != Operation::OPCODE_DEF &&
- operation.opcode != Operation::OPCODE_DEFI &&
- operation.opcode != Operation::OPCODE_DEFB)
+ if(opcode != OPCODE_DEF &&
+ opcode != OPCODE_DEFI &&
+ opcode != OPCODE_DEFB)
{
ASSERT(false);
}
}
- sourceParameter[i].swizzle = (unsigned char)((token[0] & 0x00FF0000) >> 16);
- sourceParameter[i].modifier = (SourceParameter::Modifier)((token[0] & 0x0F000000) >> 24);
+ src[i].swizzle = (unsigned char)((token[0] & 0x00FF0000) >> 16);
+ src[i].modifier = (Modifier)((token[0] & 0x0F000000) >> 24);
if((token[0] & 0x80000000) != 0x80000000)
{
- if(operation.opcode != Operation::OPCODE_DEF &&
- operation.opcode != Operation::OPCODE_DEFI &&
- operation.opcode != Operation::OPCODE_DEFB)
+ if(opcode != OPCODE_DEF &&
+ opcode != OPCODE_DEFI &&
+ opcode != OPCODE_DEFB)
{
ASSERT(false);
}
}
- if(sourceParameter[i].relative && majorVersion >= 2)
+ if(relative && majorVersion >= 2)
{
- sourceParameter[i].relativeType = (Parameter::Type)(((token[1] & 0x00001800) >> 8) | ((token[1] & 0x70000000) >> 28));
- sourceParameter[i].relativeSwizzle = (unsigned char)((token[1] & 0x00FF0000) >> 16);
+ src[i].rel.type = (ParameterType)(((token[1] & 0x00001800) >> 8) | ((token[1] & 0x70000000) >> 28));
+ src[i].rel.swizzle = (unsigned char)((token[1] & 0x00FF0000) >> 16);
}
}
}
- std::string Shader::Instruction::swizzleString(Parameter::Type type, unsigned char swizzle)
+ std::string Shader::Instruction::swizzleString(ParameterType type, unsigned char swizzle)
{
- if(type == Parameter::PARAMETER_VOID || type == Parameter::PARAMETER_LABEL || swizzle == 0xE4)
+ if(type == PARAMETER_VOID || type == PARAMETER_LABEL || swizzle == 0xE4)
{
return "";
}
@@ -803,32 +725,230 @@
return swizzleString;
}
- std::string Shader::Instruction::Parameter::string(ShaderType shaderType, unsigned short version) const
+ std::string Shader::Instruction::operationString(unsigned short version) const
{
- std::ostringstream buffer;
-
- if(type == PARAMETER_FLOATLITERAL)
+ switch(opcode)
{
- buffer << value;
-
- return buffer.str();
+ case OPCODE_NOP: return "nop";
+ case OPCODE_MOV: return "mov";
+ case OPCODE_ADD: return "add";
+ case OPCODE_SUB: return "sub";
+ case OPCODE_MAD: return "mad";
+ case OPCODE_MUL: return "mul";
+ case OPCODE_RCPX: return "rcpx";
+ case OPCODE_DIV: return "div";
+ case OPCODE_MOD: return "mod";
+ case OPCODE_RSQX: return "rsqx";
+ case OPCODE_SQRT: return "sqrt";
+ case OPCODE_RSQ: return "rsq";
+ case OPCODE_LEN2: return "len2";
+ case OPCODE_LEN3: return "len3";
+ case OPCODE_LEN4: return "len4";
+ case OPCODE_DIST1: return "dist1";
+ case OPCODE_DIST2: return "dist2";
+ case OPCODE_DIST3: return "dist3";
+ case OPCODE_DIST4: return "dist4";
+ case OPCODE_DP3: return "dp3";
+ case OPCODE_DP4: return "dp4";
+ case OPCODE_MIN: return "min";
+ case OPCODE_MAX: return "max";
+ case OPCODE_SLT: return "slt";
+ case OPCODE_SGE: return "sge";
+ case OPCODE_EXP2X: return "exp2x";
+ case OPCODE_LOG2X: return "log2x";
+ case OPCODE_LIT: return "lit";
+ case OPCODE_ATT: return "att";
+ case OPCODE_LRP: return "lrp";
+ case OPCODE_STEP: return "step";
+ case OPCODE_SMOOTH: return "smooth";
+ case OPCODE_FRC: return "frc";
+ case OPCODE_M4X4: return "m4x4";
+ case OPCODE_M4X3: return "m4x3";
+ case OPCODE_M3X4: return "m3x4";
+ case OPCODE_M3X3: return "m3x3";
+ case OPCODE_M3X2: return "m3x2";
+ case OPCODE_CALL: return "call";
+ case OPCODE_CALLNZ: return "callnz";
+ case OPCODE_LOOP: return "loop";
+ case OPCODE_RET: return "ret";
+ case OPCODE_ENDLOOP: return "endloop";
+ case OPCODE_LABEL: return "label";
+ case OPCODE_DCL: return "dcl";
+ case OPCODE_POWX: return "powx";
+ case OPCODE_CRS: return "crs";
+ case OPCODE_SGN: return "sgn";
+ case OPCODE_ABS: return "abs";
+ case OPCODE_NRM2: return "nrm2";
+ case OPCODE_NRM3: return "nrm3";
+ case OPCODE_NRM4: return "nrm4";
+ case OPCODE_SINCOS: return "sincos";
+ case OPCODE_REP: return "rep";
+ case OPCODE_ENDREP: return "endrep";
+ case OPCODE_IF: return "if";
+ case OPCODE_IFC: return "ifc";
+ case OPCODE_ELSE: return "else";
+ case OPCODE_ENDIF: return "endif";
+ case OPCODE_BREAK: return "break";
+ case OPCODE_BREAKC: return "breakc";
+ case OPCODE_MOVA: return "mova";
+ case OPCODE_DEFB: return "defb";
+ case OPCODE_DEFI: return "defi";
+ case OPCODE_TEXCOORD: return "texcoord";
+ case OPCODE_TEXKILL: return "texkill";
+ case OPCODE_DISCARD: return "discard";
+ case OPCODE_TEX:
+ if(version < 0x0104) return "tex";
+ else return "texld";
+ case OPCODE_TEXBEM: return "texbem";
+ case OPCODE_TEXBEML: return "texbeml";
+ case OPCODE_TEXREG2AR: return "texreg2ar";
+ case OPCODE_TEXREG2GB: return "texreg2gb";
+ case OPCODE_TEXM3X2PAD: return "texm3x2pad";
+ case OPCODE_TEXM3X2TEX: return "texm3x2tex";
+ case OPCODE_TEXM3X3PAD: return "texm3x3pad";
+ case OPCODE_TEXM3X3TEX: return "texm3x3tex";
+ case OPCODE_RESERVED0: return "reserved0";
+ case OPCODE_TEXM3X3SPEC: return "texm3x3spec";
+ case OPCODE_TEXM3X3VSPEC: return "texm3x3vspec";
+ case OPCODE_EXPP: return "expp";
+ case OPCODE_LOGP: return "logp";
+ case OPCODE_CND: return "cnd";
+ case OPCODE_DEF: return "def";
+ case OPCODE_TEXREG2RGB: return "texreg2rgb";
+ case OPCODE_TEXDP3TEX: return "texdp3tex";
+ case OPCODE_TEXM3X2DEPTH: return "texm3x2depth";
+ case OPCODE_TEXDP3: return "texdp3";
+ case OPCODE_TEXM3X3: return "texm3x3";
+ case OPCODE_TEXDEPTH: return "texdepth";
+ case OPCODE_CMP0: return "cmp0";
+ case OPCODE_ICMP: return "icmp";
+ case OPCODE_SELECT: return "select";
+ case OPCODE_EXTRACT: return "extract";
+ case OPCODE_INSERT: return "insert";
+ case OPCODE_BEM: return "bem";
+ case OPCODE_DP2ADD: return "dp2add";
+ case OPCODE_DFDX: return "dFdx";
+ case OPCODE_DFDY: return "dFdy";
+ case OPCODE_FWIDTH: return "fwidth";
+ case OPCODE_TEXLDD: return "texldd";
+ case OPCODE_CMP: return "cmp";
+ case OPCODE_TEXLDL: return "texldl";
+ case OPCODE_BREAKP: return "breakp";
+ case OPCODE_PHASE: return "phase";
+ case OPCODE_COMMENT: return "comment";
+ case OPCODE_END: return "end";
+ case OPCODE_PS_1_0: return "ps_1_0";
+ case OPCODE_PS_1_1: return "ps_1_1";
+ case OPCODE_PS_1_2: return "ps_1_2";
+ case OPCODE_PS_1_3: return "ps_1_3";
+ case OPCODE_PS_1_4: return "ps_1_4";
+ case OPCODE_PS_2_0: return "ps_2_0";
+ case OPCODE_PS_2_x: return "ps_2_x";
+ case OPCODE_PS_3_0: return "ps_3_0";
+ case OPCODE_VS_1_0: return "vs_1_0";
+ case OPCODE_VS_1_1: return "vs_1_1";
+ case OPCODE_VS_2_0: return "vs_2_0";
+ case OPCODE_VS_2_x: return "vs_2_x";
+ case OPCODE_VS_2_sw: return "vs_2_sw";
+ case OPCODE_VS_3_0: return "vs_3_0";
+ case OPCODE_VS_3_sw: return "vs_3_sw";
+ case OPCODE_WHILE: return "while";
+ case OPCODE_ENDWHILE: return "endwhile";
+ case OPCODE_COS: return "cos";
+ case OPCODE_SIN: return "sin";
+ case OPCODE_TAN: return "tan";
+ case OPCODE_ACOS: return "acos";
+ case OPCODE_ASIN: return "asin";
+ case OPCODE_ATAN: return "atan";
+ case OPCODE_ATAN2: return "atan2";
+ case OPCODE_DP1: return "dp1";
+ case OPCODE_DP2: return "dp2";
+ case OPCODE_TRUNC: return "trunc";
+ case OPCODE_FLOOR: return "floor";
+ case OPCODE_CEIL: return "ceil";
+ case OPCODE_EXP2: return "exp2";
+ case OPCODE_LOG2: return "log2";
+ case OPCODE_EXP: return "exp";
+ case OPCODE_LOG: return "log";
+ case OPCODE_POW: return "pow";
+ case OPCODE_F2B: return "f2b";
+ case OPCODE_B2F: return "b2f";
+ case OPCODE_ALL: return "all";
+ case OPCODE_ANY: return "any";
+ case OPCODE_NOT: return "not";
+ case OPCODE_OR: return "or";
+ case OPCODE_XOR: return "xor";
+ case OPCODE_AND: return "and";
+ case OPCODE_FORWARD1: return "forward1";
+ case OPCODE_FORWARD2: return "forward2";
+ case OPCODE_FORWARD3: return "forward3";
+ case OPCODE_FORWARD4: return "forward4";
+ case OPCODE_REFLECT1: return "reflect1";
+ case OPCODE_REFLECT2: return "reflect2";
+ case OPCODE_REFLECT3: return "reflect3";
+ case OPCODE_REFLECT4: return "reflect4";
+ case OPCODE_REFRACT1: return "refract1";
+ case OPCODE_REFRACT2: return "refract2";
+ case OPCODE_REFRACT3: return "refract3";
+ case OPCODE_REFRACT4: return "refract4";
+ case OPCODE_LEAVE: return "leave";
+ case OPCODE_CONTINUE: return "continue";
+ case OPCODE_TEST: return "test";
+ default:
+ ASSERT(false);
}
- else
- {
- if(type != PARAMETER_RASTOUT && !(type == PARAMETER_ADDR && shaderType == SHADER_VERTEX) && type != PARAMETER_LOOP && type != PARAMETER_PREDICATE && type != PARAMETER_MISCTYPE)
- {
- buffer << index;
- return typeString(shaderType, version) + buffer.str();
- }
- else
- {
- return typeString(shaderType, version);
- }
+ return "<unknown>";
+ }
+
+ std::string Shader::Instruction::controlString() const
+ {
+ if(opcode != OPCODE_LOOP && opcode != OPCODE_BREAKC && opcode != OPCODE_IFC && opcode != OPCODE_CMP)
+ {
+ if(project) return "p";
+
+ if(bias) return "b";
+
+ // FIXME: LOD
+ }
+
+ switch(control)
+ {
+ case 1: return "_gt";
+ case 2: return "_eq";
+ case 3: return "_ge";
+ case 4: return "_lt";
+ case 5: return "_ne";
+ case 6: return "_le";
+ default:
+ return "";
+ // ASSERT(false); // FIXME
}
}
- std::string Shader::Instruction::Parameter::typeString(ShaderType shaderType, unsigned short version) const
+ std::string Shader::Parameter::string(ShaderType shaderType, unsigned short version) const
+ {
+ std::ostringstream buffer;
+
+ if(type == PARAMETER_FLOAT4LITERAL)
+ {
+ buffer << '{' << value[0] << ", " << value[1] << ", " << value[2] << ", " << value[3] << '}';
+
+ return buffer.str();
+ }
+ else if(type != PARAMETER_RASTOUT && !(type == PARAMETER_ADDR && shaderType == SHADER_VERTEX) && type != PARAMETER_LOOP && type != PARAMETER_PREDICATE && type != PARAMETER_MISCTYPE)
+ {
+ buffer << index;
+
+ return typeString(shaderType, version) + buffer.str();
+ }
+ else
+ {
+ return typeString(shaderType, version);
+ }
+ }
+
+ std::string Shader::Parameter::typeString(ShaderType shaderType, unsigned short version) const
{
switch(type)
{
@@ -865,9 +985,9 @@
else ASSERT(false);
case PARAMETER_LABEL: return "l";
case PARAMETER_PREDICATE: return "p0";
- case PARAMETER_FLOATLITERAL: return "";
- case PARAMETER_BOOLLITERAL: return "";
- case PARAMETER_INTLITERAL: return "";
+ case PARAMETER_FLOAT4LITERAL: return "";
+ case PARAMETER_BOOL1LITERAL: return "";
+ case PARAMETER_INT4LITERAL: return "";
// case PARAMETER_VOID: return "";
default:
ASSERT(false);
@@ -876,55 +996,83 @@
return "";
}
- Shader::Shader(const unsigned long *shaderToken)
+ bool Shader::Instruction::isBranch() const
{
- instruction = 0;
- length = 0;
+ return opcode == OPCODE_IF || opcode == OPCODE_IFC;
+ }
+
+ bool Shader::Instruction::isCall() const
+ {
+ return opcode == OPCODE_CALL || opcode == OPCODE_CALLNZ;
+ }
- tokenCount = 0;
+ bool Shader::Instruction::isBreak() const
+ {
+ return opcode == OPCODE_BREAK || opcode == OPCODE_BREAKC || opcode == OPCODE_BREAKP;
+ }
- while(shaderToken[tokenCount] != 0x0000FFFF)
- {
- tokenCount += sw::Shader::size(shaderToken[tokenCount], (unsigned short)(shaderToken[0] & 0xFFFF)) + 1;
- }
+ bool Shader::Instruction::isLoop() const
+ {
+ return opcode == OPCODE_LOOP || opcode == OPCODE_REP || opcode == OPCODE_WHILE;
+ }
- tokenCount += 1;
+ bool Shader::Instruction::isEndLoop() const
+ {
+ return opcode == OPCODE_ENDLOOP || opcode == OPCODE_ENDREP || opcode == OPCODE_ENDWHILE;
+ }
- this->shaderToken = new unsigned long[tokenCount];
- memcpy(this->shaderToken, shaderToken, tokenCount * sizeof(unsigned long));
-
- unsigned long *hashTokens = new unsigned long[tokenCount];
- memcpy(hashTokens, shaderToken, tokenCount * sizeof(unsigned long));
- removeComments(hashTokens, tokenCount);
- hash = FNV_1((unsigned char*)hashTokens, tokenCount * sizeof(unsigned long));
- delete[] hashTokens;
+ Shader::Shader() : serialID(serialCounter++)
+ {
+ usedSamplers = 0;
}
Shader::~Shader()
{
- delete[] shaderToken;
- shaderToken = 0;
-
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
delete instruction[i];
instruction[i] = 0;
}
-
- delete[] instruction;
- instruction = 0;
}
- void Shader::getFunction(void *data, unsigned int *size)
+ void Shader::parse(const unsigned long *token)
{
- if(data)
+ minorVersion = (unsigned char)(token[0] & 0x000000FF);
+ majorVersion = (unsigned char)((token[0] & 0x0000FF00) >> 8);
+ shaderType = (ShaderType)((token[0] & 0xFFFF0000) >> 16);
+
+ int length;
+
+ if(shaderType == SHADER_VERTEX)
{
- memcpy(data, shaderToken, tokenCount * 4);
+ length = VertexShader::validate(token);
}
+ else if(shaderType == SHADER_PIXEL)
+ {
+ length = PixelShader::validate(token);
+ }
+ else ASSERT(false);
- *size = tokenCount * 4;
+ ASSERT(length != 0);
+ instruction.resize(length);
+
+ for(int i = 0; i < length; i++)
+ {
+ while((*token & 0x0000FFFF) == 0x0000FFFE) // Comment token
+ {
+ int length = (*token & 0x7FFF0000) >> 16;
+
+ token += length + 1;
+ }
+
+ int tokenCount = size(*token);
+
+ instruction[i] = new Instruction(token, tokenCount, majorVersion);
+
+ token += 1 + tokenCount;
+ }
}
-
+
int Shader::size(unsigned long opcode) const
{
return size(opcode, version);
@@ -1056,28 +1204,28 @@
int length = 0;
- if((opcode & 0x0000FFFF) == ShaderOperation::OPCODE_COMMENT)
+ if((opcode & 0x0000FFFF) == OPCODE_COMMENT)
{
return (opcode & 0x7FFF0000) >> 16;
}
- if(opcode != ShaderOperation::OPCODE_PS_1_0 &&
- opcode != ShaderOperation::OPCODE_PS_1_1 &&
- opcode != ShaderOperation::OPCODE_PS_1_2 &&
- opcode != ShaderOperation::OPCODE_PS_1_3 &&
- opcode != ShaderOperation::OPCODE_PS_1_4 &&
- opcode != ShaderOperation::OPCODE_PS_2_0 &&
- opcode != ShaderOperation::OPCODE_PS_2_x &&
- opcode != ShaderOperation::OPCODE_PS_3_0 &&
- opcode != ShaderOperation::OPCODE_VS_1_0 &&
- opcode != ShaderOperation::OPCODE_VS_1_1 &&
- opcode != ShaderOperation::OPCODE_VS_2_0 &&
- opcode != ShaderOperation::OPCODE_VS_2_x &&
- opcode != ShaderOperation::OPCODE_VS_2_sw &&
- opcode != ShaderOperation::OPCODE_VS_3_0 &&
- opcode != ShaderOperation::OPCODE_VS_3_sw &&
- opcode != ShaderOperation::OPCODE_PHASE &&
- opcode != ShaderOperation::OPCODE_END)
+ if(opcode != OPCODE_PS_1_0 &&
+ opcode != OPCODE_PS_1_1 &&
+ opcode != OPCODE_PS_1_2 &&
+ opcode != OPCODE_PS_1_3 &&
+ opcode != OPCODE_PS_1_4 &&
+ opcode != OPCODE_PS_2_0 &&
+ opcode != OPCODE_PS_2_x &&
+ opcode != OPCODE_PS_3_0 &&
+ opcode != OPCODE_VS_1_0 &&
+ opcode != OPCODE_VS_1_1 &&
+ opcode != OPCODE_VS_2_0 &&
+ opcode != OPCODE_VS_2_x &&
+ opcode != OPCODE_VS_2_sw &&
+ opcode != OPCODE_VS_3_0 &&
+ opcode != OPCODE_VS_3_sw &&
+ opcode != OPCODE_PHASE &&
+ opcode != OPCODE_END)
{
if(version >= 0x0200)
{
@@ -1098,10 +1246,10 @@
{
switch(opcode & 0x0000FFFF)
{
- case ShaderOperation::OPCODE_TEX:
+ case OPCODE_TEX:
length += 1;
break;
- case ShaderOperation::OPCODE_TEXCOORD:
+ case OPCODE_TEXCOORD:
length += 1;
break;
default:
@@ -1142,19 +1290,34 @@
return dynamicBranching;
}
- bool Shader::usesSampler(int index) const
+ bool Shader::containsBreakInstruction() const
{
- return (sampler & (1 << index)) != 0;
+ return containsBreak;
}
- int64_t Shader::getHash() const
+ bool Shader::containsContinueInstruction() const
{
- return hash;
+ return containsContinue;
+ }
+
+ bool Shader::containsLeaveInstruction() const
+ {
+ return containsLeave;
+ }
+
+ bool Shader::usesSampler(int index) const
+ {
+ return (usedSamplers & (1 << index)) != 0;
+ }
+
+ int Shader::getSerialID() const
+ {
+ return serialID;
}
int Shader::getLength() const
{
- return length;
+ return instruction.size();
}
Shader::ShaderType Shader::getShaderType() const
@@ -1176,9 +1339,9 @@
vsnprintf(fullName, 1024, fileName, vararg);
va_end(vararg);
- std::ofstream file(fullName, std::ofstream::out | std::ofstream::app);
+ std::ofstream file(fullName, std::ofstream::out);
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
file << instruction[i]->string(shaderType, version) << std::endl;
}
@@ -1191,12 +1354,19 @@
file << instruction[index]->string(shaderType, version) << std::endl;
}
- const ShaderInstruction *Shader::getInstruction(int i) const
+ void Shader::append(Instruction *instruction)
{
- if(i < 0 || i >= length)
- {
- ASSERT(false);
- }
+ this->instruction.push_back(instruction);
+ }
+
+ void Shader::declareSampler(int i)
+ {
+ usedSamplers |= 1 << i;
+ }
+
+ const Shader::Instruction *Shader::getInstruction(unsigned int i) const
+ {
+ ASSERT(i < instruction.size());
return instruction[i];
}
@@ -1207,26 +1377,26 @@
dirtyConstantsI = 0;
dirtyConstantsB = 0;
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- switch(instruction[i]->operation.opcode)
+ switch(instruction[i]->opcode)
{
- case ShaderOperation::OPCODE_DEF:
- if(instruction[i]->destinationParameter.index + 1 > dirtyConstantsF)
+ case OPCODE_DEF:
+ if(instruction[i]->dst.index + 1 > dirtyConstantsF)
{
- dirtyConstantsF = instruction[i]->destinationParameter.index + 1;
+ dirtyConstantsF = instruction[i]->dst.index + 1;
}
break;
- case ShaderOperation::OPCODE_DEFI:
- if(instruction[i]->destinationParameter.index + 1 > dirtyConstantsI)
+ case OPCODE_DEFI:
+ if(instruction[i]->dst.index + 1 > dirtyConstantsI)
{
- dirtyConstantsI = instruction[i]->destinationParameter.index + 1;
+ dirtyConstantsI = instruction[i]->dst.index + 1;
}
break;
- case ShaderOperation::OPCODE_DEFB:
- if(instruction[i]->destinationParameter.index + 1 > dirtyConstantsB)
+ case OPCODE_DEFB:
+ if(instruction[i]->dst.index + 1 > dirtyConstantsB)
{
- dirtyConstantsB = instruction[i]->destinationParameter.index + 1;
+ dirtyConstantsB = instruction[i]->dst.index + 1;
}
break;
}
@@ -1236,61 +1406,205 @@
void Shader::analyzeDynamicBranching()
{
dynamicBranching = false;
+ containsLeave = false;
+ containsBreak = false;
+ containsContinue = false;
- for(int i = 0; i < length; i++)
+ // Determine global presence of branching instructions
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- switch(instruction[i]->getOpcode())
+ switch(instruction[i]->opcode)
{
- case ShaderOperation::OPCODE_CALLNZ:
- case ShaderOperation::OPCODE_IF:
- case ShaderOperation::OPCODE_IFC:
- case ShaderOperation::OPCODE_BREAK:
- case ShaderOperation::OPCODE_BREAKC:
- case ShaderOperation::OPCODE_SETP:
- case ShaderOperation::OPCODE_BREAKP:
- if(instruction[i]->sourceParameter[0].type != ShaderParameter::PARAMETER_CONSTBOOL)
+ case OPCODE_CALLNZ:
+ case OPCODE_IF:
+ case OPCODE_IFC:
+ case OPCODE_BREAK:
+ case OPCODE_BREAKC:
+ case OPCODE_CMP:
+ case OPCODE_BREAKP:
+ case OPCODE_LEAVE:
+ case OPCODE_CONTINUE:
+ if(instruction[i]->src[0].type != PARAMETER_CONSTBOOL)
{
dynamicBranching = true;
+ }
+
+ if(instruction[i]->opcode == OPCODE_LEAVE)
+ {
+ containsLeave = true;
+ }
+
+ if(instruction[i]->isBreak())
+ {
+ containsBreak = true;
+ }
+
+ if(instruction[i]->opcode == OPCODE_CONTINUE)
+ {
+ containsContinue = true;
+ }
+ }
+ }
+
+ // Conservatively determine which instructions are affected by dynamic branching
+ int branchDepth = 0;
+ int breakDepth = 0;
+ int continueDepth = 0;
+ bool leaveReturn = false;
+
+ for(unsigned int i = 0; i < instruction.size(); i++)
+ {
+ // If statements
+ if(instruction[i]->isBranch())
+ {
+ branchDepth++;
+ }
+ else if(instruction[i]->opcode == OPCODE_ENDIF)
+ {
+ branchDepth--;
+ }
+
+ if(branchDepth > 0)
+ {
+ instruction[i]->analysisBranch = true;
+
+ if(instruction[i]->isCall())
+ {
+ markFunctionAnalysis(instruction[i]->dst.label, ANALYSIS_BRANCH);
+ }
+ }
+
+ // Break statemement
+ if(instruction[i]->isBreak())
+ {
+ breakDepth++;
+ }
+ else if(instruction[i]->isEndLoop())
+ {
+ breakDepth--;
+ }
+
+ if(breakDepth > 0)
+ {
+ if(instruction[i]->isLoop()) // Nested loop, don't make the end of it disable the break execution mask
+ {
+ breakDepth++;
+ }
+
+ instruction[i]->analysisBreak = true;
+
+ if(instruction[i]->isCall())
+ {
+ markFunctionAnalysis(instruction[i]->dst.label, ANALYSIS_BRANCH);
+ }
+ }
+
+ // Continue statement
+ if(instruction[i]->opcode == OPCODE_CONTINUE)
+ {
+ continueDepth++;
+ }
+ else if(instruction[i]->isEndLoop())
+ {
+ continueDepth--;
+ }
+
+ if(continueDepth > 0)
+ {
+ if(instruction[i]->isLoop()) // Nested loop, don't make the end of it disable the break execution mask
+ {
+ continueDepth++;
+ }
+
+ instruction[i]->analysisContinue = true;
+
+ if(instruction[i]->isCall())
+ {
+ markFunctionAnalysis(instruction[i]->dst.label, ANALYSIS_CONTINUE);
+ }
+ }
+
+ // Return (leave) statement
+ if(instruction[i]->opcode == OPCODE_LEAVE)
+ {
+ leaveReturn = true;
+ }
+ else if(instruction[i]->opcode == OPCODE_RET) // End of the function
+ {
+ leaveReturn = false;
+ }
+
+ if(leaveReturn)
+ {
+ instruction[i]->analysisLeave = true;
+
+ if(instruction[i]->isCall())
+ {
+ markFunctionAnalysis(instruction[i]->dst.label, ANALYSIS_LEAVE);
+ }
+ }
+ }
+ }
+
+ void Shader::markFunctionAnalysis(int functionLabel, Analysis flag)
+ {
+ bool marker = false;
+ for(unsigned int i = 0; i < instruction.size(); i++)
+ {
+ if(!marker)
+ {
+ if(instruction[i]->opcode == OPCODE_LABEL && instruction[i]->dst.label == functionLabel)
+ {
+ marker = true;
+ }
+ }
+ else
+ {
+ if(instruction[i]->opcode == OPCODE_RET)
+ {
break;
}
+ else if(instruction[i]->isCall())
+ {
+ markFunctionAnalysis(instruction[i]->dst.label, flag);
+ }
+
+ instruction[i]->analysis |= flag;
}
}
}
void Shader::analyzeSamplers()
{
- sampler = 0;
-
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- switch(instruction[i]->getOpcode())
+ switch(instruction[i]->opcode)
{
- case ShaderOperation::OPCODE_TEX:
- case ShaderOperation::OPCODE_TEXBEM:
- case ShaderOperation::OPCODE_TEXBEML:
- case ShaderOperation::OPCODE_TEXREG2AR:
- case ShaderOperation::OPCODE_TEXREG2GB:
- case ShaderOperation::OPCODE_TEXM3X2TEX:
- case ShaderOperation::OPCODE_TEXM3X3TEX:
- case ShaderOperation::OPCODE_TEXM3X3SPEC:
- case ShaderOperation::OPCODE_TEXM3X3VSPEC:
- case ShaderOperation::OPCODE_TEXREG2RGB:
- case ShaderOperation::OPCODE_TEXDP3TEX:
- case ShaderOperation::OPCODE_TEXM3X2DEPTH:
- case ShaderOperation::OPCODE_TEXLDD:
- case ShaderOperation::OPCODE_TEXLDL:
+ case OPCODE_TEX:
+ case OPCODE_TEXBEM:
+ case OPCODE_TEXBEML:
+ case OPCODE_TEXREG2AR:
+ case OPCODE_TEXREG2GB:
+ case OPCODE_TEXM3X2TEX:
+ case OPCODE_TEXM3X3TEX:
+ case OPCODE_TEXM3X3SPEC:
+ case OPCODE_TEXM3X3VSPEC:
+ case OPCODE_TEXREG2RGB:
+ case OPCODE_TEXDP3TEX:
+ case OPCODE_TEXM3X2DEPTH:
+ case OPCODE_TEXLDD:
+ case OPCODE_TEXLDL:
{
- ShaderParameter &dst = instruction[i]->destinationParameter;
- ShaderParameter &src1 = instruction[i]->sourceParameter[1];
+ Parameter &dst = instruction[i]->dst;
+ Parameter &src1 = instruction[i]->src[1];
if(majorVersion >= 2)
{
- ASSERT(src1.type == ShaderParameter::PARAMETER_SAMPLER);
- sampler |= 1 << src1.index;
+ usedSamplers |= 1 << src1.index;
}
else
{
- sampler |= 1 << dst.index;
+ usedSamplers |= 1 << dst.index;
}
}
break;
@@ -1298,21 +1612,57 @@
}
}
- void Shader::removeComments(unsigned long *shaderToken, int tokenCount)
+ // Assigns a unique index to each call instruction, on a per label basis.
+ // This is used to know what basic block to return to.
+ void Shader::analyzeCallSites()
{
- for(int i = 0; i < tokenCount; )
- {
- int instructionSize = sw::Shader::size(shaderToken[i], (unsigned short)(shaderToken[0] & 0xFFFF)) + 1;
+ int callSiteIndex[2048] = {0};
- if((shaderToken[i] & 0x0000FFFF) == ShaderOperation::OPCODE_COMMENT)
+ for(unsigned int i = 0; i < instruction.size(); i++)
+ {
+ if(instruction[i]->opcode == OPCODE_CALL || instruction[i]->opcode == OPCODE_CALLNZ)
{
- for(int j = 0; j < instructionSize; j++)
+ int label = instruction[i]->dst.label;
+
+ instruction[i]->dst.callSite = callSiteIndex[label]++;
+ }
+ }
+ }
+
+ void Shader::analyzeDynamicIndexing()
+ {
+ dynamicallyIndexedTemporaries = false;
+ dynamicallyIndexedInput = false;
+ dynamicallyIndexedOutput = false;
+
+ for(unsigned int i = 0; i < instruction.size(); i++)
+ {
+ if(instruction[i]->dst.rel.type == PARAMETER_ADDR ||
+ instruction[i]->dst.rel.type == PARAMETER_LOOP ||
+ instruction[i]->dst.rel.type == PARAMETER_TEMP)
+ {
+ switch(instruction[i]->dst.type)
{
- shaderToken[i + j] = ShaderOperation::OPCODE_NOP;
+ case PARAMETER_TEMP: dynamicallyIndexedTemporaries = true; break;
+ case PARAMETER_INPUT: dynamicallyIndexedInput = true; break;
+ case PARAMETER_OUTPUT: dynamicallyIndexedOutput = true; break;
}
}
- i += instructionSize;
+ for(int j = 0; j < 3; j++)
+ {
+ if(instruction[i]->src[j].rel.type == PARAMETER_ADDR ||
+ instruction[i]->src[j].rel.type == PARAMETER_LOOP ||
+ instruction[i]->src[j].rel.type == PARAMETER_TEMP)
+ {
+ switch(instruction[i]->src[j].type)
+ {
+ case PARAMETER_TEMP: dynamicallyIndexedTemporaries = true; break;
+ case PARAMETER_INPUT: dynamicallyIndexedInput = true; break;
+ case PARAMETER_OUTPUT: dynamicallyIndexedOutput = true; break;
+ }
+ }
+ }
}
}
}
diff --git a/src/Shader/Shader.hpp b/src/Shader/Shader.hpp
index d2d6fe3..b08b56e 100644
--- a/src/Shader/Shader.hpp
+++ b/src/Shader/Shader.hpp
@@ -1,459 +1,572 @@
-// SwiftShader Software Renderer
-//
-// Copyright(c) 2005-2011 TransGaming Inc.
-//
-// All rights reserved. No part of this software may be copied, distributed, transmitted,
-// transcribed, stored in a retrieval system, translated into any human or computer
-// language by any means, or disclosed to third parties without the explicit written
-// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
-// or implied, including but not limited to any patent rights, are granted to you.
-//
-
-#ifndef sw_Shader_hpp
-#define sw_Shader_hpp
-
-#include "Common/Types.hpp"
-
-#include <string>
-
-namespace sw
-{
- class Shader
- {
- public:
- enum ShaderType
- {
- SHADER_PIXEL = 0xFFFF,
- SHADER_VERTEX = 0xFFFE,
- SHADER_GEOMETRY = 0xFFFD
- };
-
- class Instruction
- {
- friend Shader;
-
- public:
- Instruction();
- Instruction(const unsigned long *token, int size, unsigned char majorVersion);
-
- virtual ~Instruction();
-
- struct Operation
- {
- enum Opcode
- {
- // Extracted from d3d9types.h
- OPCODE_NOP = 0,
- OPCODE_MOV,
- OPCODE_ADD,
- OPCODE_SUB,
- OPCODE_MAD,
- OPCODE_MUL,
- OPCODE_RCP,
- OPCODE_RSQ,
- OPCODE_DP3,
- OPCODE_DP4,
- OPCODE_MIN,
- OPCODE_MAX,
- OPCODE_SLT,
- OPCODE_SGE,
- OPCODE_EXP,
- OPCODE_LOG,
- OPCODE_LIT,
- OPCODE_DST,
- OPCODE_LRP,
- OPCODE_FRC,
- OPCODE_M4X4,
- OPCODE_M4X3,
- OPCODE_M3X4,
- OPCODE_M3X3,
- OPCODE_M3X2,
- OPCODE_CALL,
- OPCODE_CALLNZ,
- OPCODE_LOOP,
- OPCODE_RET,
- OPCODE_ENDLOOP,
- OPCODE_LABEL,
- OPCODE_DCL,
- OPCODE_POW,
- OPCODE_CRS,
- OPCODE_SGN,
- OPCODE_ABS,
- OPCODE_NRM,
- OPCODE_SINCOS,
- OPCODE_REP,
- OPCODE_ENDREP,
- OPCODE_IF,
- OPCODE_IFC,
- OPCODE_ELSE,
- OPCODE_ENDIF,
- OPCODE_BREAK,
- OPCODE_BREAKC,
- OPCODE_MOVA,
- OPCODE_DEFB,
- OPCODE_DEFI,
-
- OPCODE_TEXCOORD = 64,
- OPCODE_TEXKILL,
- OPCODE_TEX,
- OPCODE_TEXBEM,
- OPCODE_TEXBEML,
- OPCODE_TEXREG2AR,
- OPCODE_TEXREG2GB,
- OPCODE_TEXM3X2PAD,
- OPCODE_TEXM3X2TEX,
- OPCODE_TEXM3X3PAD,
- OPCODE_TEXM3X3TEX,
- OPCODE_RESERVED0,
- OPCODE_TEXM3X3SPEC,
- OPCODE_TEXM3X3VSPEC,
- OPCODE_EXPP,
- OPCODE_LOGP,
- OPCODE_CND,
- OPCODE_DEF,
- OPCODE_TEXREG2RGB,
- OPCODE_TEXDP3TEX,
- OPCODE_TEXM3X2DEPTH,
- OPCODE_TEXDP3,
- OPCODE_TEXM3X3,
- OPCODE_TEXDEPTH,
- OPCODE_CMP,
- OPCODE_BEM,
- OPCODE_DP2ADD,
- OPCODE_DSX,
- OPCODE_DSY,
- OPCODE_TEXLDD,
- OPCODE_SETP,
- OPCODE_TEXLDL,
- OPCODE_BREAKP,
-
- OPCODE_PHASE = 0xFFFD,
- OPCODE_COMMENT = 0xFFFE,
- OPCODE_END = 0xFFFF,
-
- OPCODE_PS_1_0 = 0xFFFF0100,
- OPCODE_PS_1_1 = 0xFFFF0101,
- OPCODE_PS_1_2 = 0xFFFF0102,
- OPCODE_PS_1_3 = 0xFFFF0103,
- OPCODE_PS_1_4 = 0xFFFF0104,
- OPCODE_PS_2_0 = 0xFFFF0200,
- OPCODE_PS_2_x = 0xFFFF0201,
- OPCODE_PS_3_0 = 0xFFFF0300,
-
- OPCODE_VS_1_0 = 0xFFFE0100,
- OPCODE_VS_1_1 = 0xFFFE0101,
- OPCODE_VS_2_0 = 0xFFFE0200,
- OPCODE_VS_2_x = 0xFFFE0201,
- OPCODE_VS_2_sw = 0xFFFE02FF,
- OPCODE_VS_3_0 = 0xFFFE0300,
- OPCODE_VS_3_sw = 0xFFFE03FF,
- };
-
- enum Control
- {
- CONTROL_RESERVED0,
- CONTROL_GT,
- CONTROL_EQ,
- CONTROL_GE,
- CONTROL_LT,
- CONTROL_NE,
- CONTROL_LE,
- CONTROL_RESERVED1
- };
-
- enum SamplerType
- {
- SAMPLER_UNKNOWN,
- SAMPLER_1D,
- SAMPLER_2D,
- SAMPLER_CUBE,
- SAMPLER_VOLUME
- };
-
- enum Usage // For vertex input/output declarations
- {
- USAGE_POSITION = 0,
- USAGE_BLENDWEIGHT = 1,
- USAGE_BLENDINDICES = 2,
- USAGE_NORMAL = 3,
- USAGE_PSIZE = 4,
- USAGE_TEXCOORD = 5,
- USAGE_TANGENT = 6,
- USAGE_BINORMAL = 7,
- USAGE_TESSFACTOR = 8,
- USAGE_POSITIONT = 9,
- USAGE_COLOR = 10,
- USAGE_FOG = 11,
- USAGE_DEPTH = 12,
- USAGE_SAMPLE = 13
- };
-
- Operation() : opcode(OPCODE_NOP), control(CONTROL_RESERVED0), predicate(false), predicateNot(false), predicateSwizzle(0xE4), coissue(false), samplerType(SAMPLER_UNKNOWN), usage(USAGE_POSITION), usageIndex(0)
- {
- }
-
- std::string string(unsigned short version) const;
- std::string controlString() const;
-
- Opcode opcode;
-
- union
- {
- Control control;
-
- struct
- {
- unsigned char project : 1;
- unsigned char bias : 1;
- };
- };
-
- bool predicate;
- bool predicateNot; // Negative predicate
- unsigned char predicateSwizzle;
-
- bool coissue;
- SamplerType samplerType;
- Usage usage;
- unsigned char usageIndex;
- };
-
- struct Parameter
- {
- enum Type
- {
- PARAMETER_TEMP = 0,
- PARAMETER_INPUT = 1,
- PARAMETER_CONST = 2,
- PARAMETER_TEXTURE = 3,
- PARAMETER_ADDR = 3,
- PARAMETER_RASTOUT = 4,
- PARAMETER_ATTROUT = 5,
- PARAMETER_TEXCRDOUT = 6,
- PARAMETER_OUTPUT = 6,
- PARAMETER_CONSTINT = 7,
- PARAMETER_COLOROUT = 8,
- PARAMETER_DEPTHOUT = 9,
- PARAMETER_SAMPLER = 10,
- PARAMETER_CONST2 = 11,
- PARAMETER_CONST3 = 12,
- PARAMETER_CONST4 = 13,
- PARAMETER_CONSTBOOL = 14,
- PARAMETER_LOOP = 15,
- PARAMETER_TEMPFLOAT16 = 16,
- PARAMETER_MISCTYPE = 17,
- PARAMETER_LABEL = 18,
- PARAMETER_PREDICATE = 19,
-
- // Internally used
- PARAMETER_FLOATLITERAL = 20,
- PARAMETER_BOOLLITERAL = 21,
- PARAMETER_INTLITERAL = 22,
-
- PARAMETER_VOID
- };
-
- union
- {
- unsigned int index; // For registers
- float value; // For float constants
- int integer; // For integer constants
- bool boolean; // For boolean constants
- };
-
- Parameter() : type(PARAMETER_VOID), index(0), relative(false), relativeType(PARAMETER_VOID), relativeSwizzle(0xE4)
- {
- }
-
- std::string string(ShaderType shaderType, unsigned short version) const;
- std::string typeString(ShaderType shaderType, unsigned short version) const;
- std::string relativeString() const;
-
- Type type;
- bool relative;
- Type relativeType;
- unsigned char relativeSwizzle;
- };
-
- struct DestinationParameter : Parameter
- {
- union
- {
- unsigned char mask;
-
- struct
- {
- bool x : 1;
- bool y : 1;
- bool z : 1;
- bool w : 1;
- };
- };
-
- DestinationParameter() : mask(0xF), saturate(false), partialPrecision(false), centroid(false), shift(0)
- {
- }
-
- std::string modifierString() const;
- std::string shiftString() const;
- std::string maskString() const;
-
- bool saturate;
- bool partialPrecision;
- bool centroid;
- signed char shift;
- };
-
- struct SourceParameter : Parameter
- {
- enum Modifier
- {
- MODIFIER_NONE,
- MODIFIER_NEGATE,
- MODIFIER_BIAS,
- MODIFIER_BIAS_NEGATE,
- MODIFIER_SIGN,
- MODIFIER_SIGN_NEGATE,
- MODIFIER_COMPLEMENT,
- MODIFIER_X2,
- MODIFIER_X2_NEGATE,
- MODIFIER_DZ,
- MODIFIER_DW,
- MODIFIER_ABS,
- MODIFIER_ABS_NEGATE,
- MODIFIER_NOT
- };
-
- SourceParameter() : swizzle(0xE4), modifier(MODIFIER_NONE)
- {
- }
-
- std::string swizzleString() const;
- std::string preModifierString() const;
- std::string postModifierString() const;
-
- unsigned char swizzle;
- Modifier modifier;
- };
-
- void parseOperationToken(unsigned long token, unsigned char majorVersion);
- void parseDeclarationToken(unsigned long token);
- void parseDestinationToken(const unsigned long *token, unsigned char majorVersion);
- void parseSourceToken(int i, const unsigned long *token, unsigned char majorVersion);
-
- Operation::Opcode getOpcode() const;
- const DestinationParameter &getDestinationParameter() const;
- const SourceParameter &getSourceParameter(int i) const;
-
- bool isCoissue() const;
- bool isProject() const;
- bool isBias() const;
- bool isPredicate() const;
- bool isPredicateNot() const;
- unsigned char getPredicateSwizzle() const;
- Operation::Control getControl() const;
- Operation::Usage getUsage() const;
- unsigned char getUsageIndex() const;
- Operation::SamplerType getSamplerType() const;
-
- std::string string(ShaderType shaderType, unsigned short version) const;
-
- protected:
- Operation operation;
- DestinationParameter destinationParameter;
- SourceParameter sourceParameter[4];
-
- private:
- static std::string swizzleString(Parameter::Type type, unsigned char swizzle);
- };
-
- Shader(const unsigned long *shaderToken);
-
- ~Shader();
-
- void getFunction(void *data, unsigned int *size);
-
- int64_t getHash() const;
- int getLength() const;
- ShaderType getShaderType() const;
- unsigned short getVersion() const;
-
- const Instruction *getInstruction(int i) const;
- int size(unsigned long opcode) const;
- static int size(unsigned long opcode, unsigned short version);
-
- void print(const char *fileName, ...) const;
- void printInstruction(int index, const char *fileName) const;
-
- static bool maskContainsComponent(int mask, int component);
- static bool swizzleContainsComponent(int swizzle, int component);
- static bool swizzleContainsComponentMasked(int swizzle, int component, int mask);
-
- bool containsDynamicBranching() const;
- bool usesSampler(int i) const;
-
- struct Semantic
- {
- Semantic(unsigned char usage = 0xFF, unsigned char index = 0xFF) : usage(usage), index(index), centroid(false)
- {
- }
-
- bool operator==(const Semantic &semantic) const
- {
- return usage == semantic.usage && index == semantic.index;
- }
-
- bool active() const
- {
- return usage != 0xFF;
- }
-
- unsigned char usage;
- unsigned char index;
- bool centroid;
- };
-
- unsigned int dirtyConstantsF; // FIXME: Private
- unsigned int dirtyConstantsI; // FIXME: Private
- unsigned int dirtyConstantsB; // FIXME: Private
-
- protected:
- void analyzeDirtyConstants();
- void analyzeDynamicBranching();
- void analyzeSamplers();
-
- ShaderType shaderType;
-
- union
- {
- unsigned short version;
-
- struct
- {
- unsigned char minorVersion;
- unsigned char majorVersion;
- };
- };
-
- int length;
- Instruction **instruction;
-
- private:
- static void removeComments(unsigned long *shaderToken, int tokenCount);
-
- int64_t hash;
-
- bool dynamicBranching;
- unsigned short sampler;
-
- unsigned long *shaderToken;
- int tokenCount;
- };
-
- typedef Shader::Instruction::Operation ShaderOperation;
- typedef Shader::Instruction ShaderInstruction;
- typedef Shader::Instruction::Parameter ShaderParameter;
- typedef Shader::Instruction::Operation::Opcode ShaderOpcode;
-}
-
-#endif // sw_Shader_hpp
+// SwiftShader Software Renderer
+//
+// Copyright(c) 2005-2012 TransGaming Inc.
+//
+// All rights reserved. No part of this software may be copied, distributed, transmitted,
+// transcribed, stored in a retrieval system, translated into any human or computer
+// language by any means, or disclosed to third parties without the explicit written
+// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
+// or implied, including but not limited to any patent rights, are granted to you.
+//
+
+#ifndef sw_Shader_hpp
+#define sw_Shader_hpp
+
+#include "Common/Types.hpp"
+
+#include <string>
+#include <vector>
+
+namespace sw
+{
+ class Shader
+ {
+ public:
+ enum ShaderType
+ {
+ SHADER_PIXEL = 0xFFFF,
+ SHADER_VERTEX = 0xFFFE,
+ SHADER_GEOMETRY = 0xFFFD
+ };
+
+ enum Opcode
+ {
+ // Matches order in d3d9types.h
+ OPCODE_NOP = 0,
+ OPCODE_MOV,
+ OPCODE_ADD,
+ OPCODE_SUB,
+ OPCODE_MAD,
+ OPCODE_MUL,
+ OPCODE_RCPX,
+ OPCODE_RSQX,
+ OPCODE_DP3,
+ OPCODE_DP4,
+ OPCODE_MIN,
+ OPCODE_MAX,
+ OPCODE_SLT,
+ OPCODE_SGE,
+ OPCODE_EXP2X, // D3DSIO_EXP
+ OPCODE_LOG2X, // D3DSIO_LOG
+ OPCODE_LIT,
+ OPCODE_ATT, // D3DSIO_DST
+ OPCODE_LRP,
+ OPCODE_FRC,
+ OPCODE_M4X4,
+ OPCODE_M4X3,
+ OPCODE_M3X4,
+ OPCODE_M3X3,
+ OPCODE_M3X2,
+ OPCODE_CALL,
+ OPCODE_CALLNZ,
+ OPCODE_LOOP,
+ OPCODE_RET,
+ OPCODE_ENDLOOP,
+ OPCODE_LABEL,
+ OPCODE_DCL,
+ OPCODE_POWX,
+ OPCODE_CRS,
+ OPCODE_SGN,
+ OPCODE_ABS,
+ OPCODE_NRM3, // D3DSIO_NRM
+ OPCODE_SINCOS,
+ OPCODE_REP,
+ OPCODE_ENDREP,
+ OPCODE_IF,
+ OPCODE_IFC,
+ OPCODE_ELSE,
+ OPCODE_ENDIF,
+ OPCODE_BREAK,
+ OPCODE_BREAKC,
+ OPCODE_MOVA,
+ OPCODE_DEFB,
+ OPCODE_DEFI,
+
+ OPCODE_TEXCOORD = 64,
+ OPCODE_TEXKILL,
+ OPCODE_TEX,
+ OPCODE_TEXBEM,
+ OPCODE_TEXBEML,
+ OPCODE_TEXREG2AR,
+ OPCODE_TEXREG2GB,
+ OPCODE_TEXM3X2PAD,
+ OPCODE_TEXM3X2TEX,
+ OPCODE_TEXM3X3PAD,
+ OPCODE_TEXM3X3TEX,
+ OPCODE_RESERVED0,
+ OPCODE_TEXM3X3SPEC,
+ OPCODE_TEXM3X3VSPEC,
+ OPCODE_EXPP,
+ OPCODE_LOGP,
+ OPCODE_CND,
+ OPCODE_DEF,
+ OPCODE_TEXREG2RGB,
+ OPCODE_TEXDP3TEX,
+ OPCODE_TEXM3X2DEPTH,
+ OPCODE_TEXDP3,
+ OPCODE_TEXM3X3,
+ OPCODE_TEXDEPTH,
+ OPCODE_CMP0, // D3DSIO_CMP
+ OPCODE_BEM,
+ OPCODE_DP2ADD,
+ OPCODE_DFDX, // D3DSIO_DSX
+ OPCODE_DFDY, // D3DSIO_DSY
+ OPCODE_TEXLDD,
+ OPCODE_CMP, // D3DSIO_SETP
+ OPCODE_TEXLDL,
+ OPCODE_BREAKP,
+
+ OPCODE_PHASE = 0xFFFD,
+ OPCODE_COMMENT = 0xFFFE,
+ OPCODE_END = 0xFFFF,
+
+ OPCODE_PS_1_0 = 0xFFFF0100,
+ OPCODE_PS_1_1 = 0xFFFF0101,
+ OPCODE_PS_1_2 = 0xFFFF0102,
+ OPCODE_PS_1_3 = 0xFFFF0103,
+ OPCODE_PS_1_4 = 0xFFFF0104,
+ OPCODE_PS_2_0 = 0xFFFF0200,
+ OPCODE_PS_2_x = 0xFFFF0201,
+ OPCODE_PS_3_0 = 0xFFFF0300,
+
+ OPCODE_VS_1_0 = 0xFFFE0100,
+ OPCODE_VS_1_1 = 0xFFFE0101,
+ OPCODE_VS_2_0 = 0xFFFE0200,
+ OPCODE_VS_2_x = 0xFFFE0201,
+ OPCODE_VS_2_sw = 0xFFFE02FF,
+ OPCODE_VS_3_0 = 0xFFFE0300,
+ OPCODE_VS_3_sw = 0xFFFE03FF,
+
+ OPCODE_WHILE = 0x80000001,
+ OPCODE_ENDWHILE,
+ OPCODE_COS,
+ OPCODE_SIN,
+ OPCODE_TAN,
+ OPCODE_ACOS,
+ OPCODE_ASIN,
+ OPCODE_ATAN,
+ OPCODE_ATAN2,
+ OPCODE_DP1,
+ OPCODE_DP2,
+ OPCODE_TRUNC,
+ OPCODE_FLOOR,
+ OPCODE_CEIL,
+ OPCODE_SQRT,
+ OPCODE_RSQ,
+ OPCODE_LEN2,
+ OPCODE_LEN3,
+ OPCODE_LEN4,
+ OPCODE_DIST1,
+ OPCODE_DIST2,
+ OPCODE_DIST3,
+ OPCODE_DIST4,
+ OPCODE_NRM2,
+ OPCODE_NRM4,
+ OPCODE_DIV,
+ OPCODE_MOD,
+ OPCODE_EXP2,
+ OPCODE_LOG2,
+ OPCODE_EXP,
+ OPCODE_LOG,
+ OPCODE_POW,
+ OPCODE_F2B, // Float to bool
+ OPCODE_B2F, // Bool to float
+ OPCODE_ALL,
+ OPCODE_ANY,
+ OPCODE_NOT,
+ OPCODE_OR,
+ OPCODE_XOR,
+ OPCODE_AND,
+ OPCODE_STEP,
+ OPCODE_SMOOTH,
+ OPCODE_FORWARD1,
+ OPCODE_FORWARD2,
+ OPCODE_FORWARD3,
+ OPCODE_FORWARD4,
+ OPCODE_REFLECT1,
+ OPCODE_REFLECT2,
+ OPCODE_REFLECT3,
+ OPCODE_REFLECT4,
+ OPCODE_REFRACT1,
+ OPCODE_REFRACT2,
+ OPCODE_REFRACT3,
+ OPCODE_REFRACT4,
+ OPCODE_ICMP,
+ OPCODE_SELECT,
+ OPCODE_EXTRACT,
+ OPCODE_INSERT,
+ OPCODE_DISCARD,
+ OPCODE_FWIDTH,
+ OPCODE_LEAVE, // Return before the end of the function
+ OPCODE_CONTINUE,
+ OPCODE_TEST, // Marks the end of the code that can be skipped by 'continue'
+ };
+
+ static Opcode OPCODE_DP(int);
+ static Opcode OPCODE_LEN(int);
+ static Opcode OPCODE_DIST(int);
+ static Opcode OPCODE_NRM(int);
+ static Opcode OPCODE_FORWARD(int);
+ static Opcode OPCODE_REFLECT(int);
+ static Opcode OPCODE_REFRACT(int);
+
+ enum Control
+ {
+ CONTROL_RESERVED0,
+ CONTROL_GT,
+ CONTROL_EQ,
+ CONTROL_GE,
+ CONTROL_LT,
+ CONTROL_NE,
+ CONTROL_LE,
+ CONTROL_RESERVED1
+ };
+
+ enum SamplerType
+ {
+ SAMPLER_UNKNOWN,
+ SAMPLER_1D,
+ SAMPLER_2D,
+ SAMPLER_CUBE,
+ SAMPLER_VOLUME
+ };
+
+ enum Usage // For vertex input/output declarations
+ {
+ USAGE_POSITION = 0,
+ USAGE_BLENDWEIGHT = 1,
+ USAGE_BLENDINDICES = 2,
+ USAGE_NORMAL = 3,
+ USAGE_PSIZE = 4,
+ USAGE_TEXCOORD = 5,
+ USAGE_TANGENT = 6,
+ USAGE_BINORMAL = 7,
+ USAGE_TESSFACTOR = 8,
+ USAGE_POSITIONT = 9,
+ USAGE_COLOR = 10,
+ USAGE_FOG = 11,
+ USAGE_DEPTH = 12,
+ USAGE_SAMPLE = 13
+ };
+
+ enum ParameterType
+ {
+ PARAMETER_TEMP = 0,
+ PARAMETER_INPUT = 1,
+ PARAMETER_CONST = 2,
+ PARAMETER_TEXTURE = 3,
+ PARAMETER_ADDR = 3,
+ PARAMETER_RASTOUT = 4,
+ PARAMETER_ATTROUT = 5,
+ PARAMETER_TEXCRDOUT = 6,
+ PARAMETER_OUTPUT = 6,
+ PARAMETER_CONSTINT = 7,
+ PARAMETER_COLOROUT = 8,
+ PARAMETER_DEPTHOUT = 9,
+ PARAMETER_SAMPLER = 10,
+ PARAMETER_CONST2 = 11,
+ PARAMETER_CONST3 = 12,
+ PARAMETER_CONST4 = 13,
+ PARAMETER_CONSTBOOL = 14,
+ PARAMETER_LOOP = 15,
+ PARAMETER_TEMPFLOAT16 = 16,
+ PARAMETER_MISCTYPE = 17,
+ PARAMETER_LABEL = 18,
+ PARAMETER_PREDICATE = 19,
+
+ // PARAMETER_FLOAT1LITERAL,
+ // PARAMETER_FLOAT2LITERAL,
+ // PARAMETER_FLOAT3LITERAL,
+ PARAMETER_FLOAT4LITERAL,
+ PARAMETER_BOOL1LITERAL,
+ // PARAMETER_BOOL2LITERAL,
+ // PARAMETER_BOOL3LITERAL,
+ // PARAMETER_BOOL4LITERAL,
+ // PARAMETER_INT1LITERAL,
+ // PARAMETER_INT2LITERAL,
+ // PARAMETER_INT3LITERAL,
+ PARAMETER_INT4LITERAL,
+
+ PARAMETER_VOID
+ };
+
+ enum Modifier
+ {
+ MODIFIER_NONE,
+ MODIFIER_NEGATE,
+ MODIFIER_BIAS,
+ MODIFIER_BIAS_NEGATE,
+ MODIFIER_SIGN,
+ MODIFIER_SIGN_NEGATE,
+ MODIFIER_COMPLEMENT,
+ MODIFIER_X2,
+ MODIFIER_X2_NEGATE,
+ MODIFIER_DZ,
+ MODIFIER_DW,
+ MODIFIER_ABS,
+ MODIFIER_ABS_NEGATE,
+ MODIFIER_NOT
+ };
+
+ enum Analysis
+ {
+ // Flags indicating whether an instruction is affected by an execution enable mask
+ ANALYSIS_BRANCH = 0x00000001,
+ ANALYSIS_BREAK = 0x00000002,
+ ANALYSIS_CONTINUE = 0x00000004,
+ ANALYSIS_LEAVE = 0x00000008,
+ };
+
+ struct RelativeAddress
+ {
+ RelativeAddress() : type(PARAMETER_VOID), index(0), swizzle(0), scale(1), deterministic(false)
+ {
+ }
+
+ ParameterType type : 8;
+ unsigned int index;
+ unsigned int swizzle : 8;
+ unsigned int scale;
+ bool deterministic; // Equal accross shader instances run in lockstep (e.g. unrollable loop couters)
+ };
+
+ struct Parameter
+ {
+ union
+ {
+ struct
+ {
+ unsigned int index; // For registers types
+
+ RelativeAddress rel;
+ };
+
+ float value[4]; // For float constants
+ int integer[4]; // For integer constants
+ int boolean[4]; // For boolean constants
+
+ struct
+ {
+ unsigned int label; // Label index
+ unsigned int callSite; // Call index (per label)
+ };
+ };
+
+ Parameter() : type(PARAMETER_VOID), index(0)
+ {
+ }
+
+ std::string string(ShaderType shaderType, unsigned short version) const;
+ std::string typeString(ShaderType shaderType, unsigned short version) const;
+ std::string relativeString() const;
+
+ ParameterType type : 8;
+ };
+
+ struct DestinationParameter : Parameter
+ {
+ union
+ {
+ unsigned char mask;
+
+ struct
+ {
+ bool x : 1;
+ bool y : 1;
+ bool z : 1;
+ bool w : 1;
+ };
+ };
+
+ DestinationParameter() : mask(0xF), integer(false), saturate(false), partialPrecision(false), centroid(false), shift(0)
+ {
+ }
+
+ std::string modifierString() const;
+ std::string shiftString() const;
+ std::string maskString() const;
+
+ bool integer : 1;
+ bool saturate : 1;
+ bool partialPrecision : 1;
+ bool centroid : 1;
+ signed char shift : 4;
+ };
+
+ struct SourceParameter : Parameter
+ {
+ SourceParameter() : swizzle(0xE4), modifier(MODIFIER_NONE)
+ {
+ }
+
+ std::string swizzleString() const;
+ std::string preModifierString() const;
+ std::string postModifierString() const;
+
+ unsigned int swizzle : 8;
+ Modifier modifier : 8;
+ };
+
+ struct Instruction
+ {
+ explicit Instruction(Opcode opcode);
+ Instruction(const unsigned long *token, int size, unsigned char majorVersion);
+
+ virtual ~Instruction();
+
+ void parseOperationToken(unsigned long token, unsigned char majorVersion);
+ void parseDeclarationToken(unsigned long token);
+ void parseDestinationToken(const unsigned long *token, unsigned char majorVersion);
+ void parseSourceToken(int i, const unsigned long *token, unsigned char majorVersion);
+
+ std::string string(ShaderType shaderType, unsigned short version) const;
+ static std::string swizzleString(ParameterType type, unsigned char swizzle);
+ std::string operationString(unsigned short version) const;
+ std::string controlString() const;
+
+ bool isBranch() const;
+ bool isCall() const;
+ bool isBreak() const;
+ bool isLoop() const;
+ bool isEndLoop() const;
+
+ Opcode opcode;
+
+ union
+ {
+ Control control;
+
+ struct
+ {
+ unsigned char project : 1;
+ unsigned char bias : 1;
+ };
+ };
+
+ bool predicate;
+ bool predicateNot; // Negative predicate
+ unsigned char predicateSwizzle;
+
+ bool coissue;
+ SamplerType samplerType;
+ Usage usage;
+ unsigned char usageIndex;
+
+ DestinationParameter dst;
+ SourceParameter src[4];
+
+ union
+ {
+ unsigned int analysis;
+
+ struct
+ {
+ // Keep in sync with Shader::Analysis flags
+ unsigned int analysisBranch : 1;
+ unsigned int analysisBreak : 1;
+ unsigned int analysisContinue : 1;
+ unsigned int analysisLeave : 1;
+ };
+ };
+ };
+
+ Shader();
+
+ ~Shader();
+
+ int getSerialID() const;
+ int getLength() const;
+ ShaderType getShaderType() const;
+ unsigned short getVersion() const;
+
+ void append(Instruction *instruction);
+ void declareSampler(int i);
+
+ const Instruction *getInstruction(unsigned int i) const;
+ int size(unsigned long opcode) const;
+ static int size(unsigned long opcode, unsigned short version);
+
+ void print(const char *fileName, ...) const;
+ void printInstruction(int index, const char *fileName) const;
+
+ static bool maskContainsComponent(int mask, int component);
+ static bool swizzleContainsComponent(int swizzle, int component);
+ static bool swizzleContainsComponentMasked(int swizzle, int component, int mask);
+
+ bool containsDynamicBranching() const;
+ bool containsBreakInstruction() const;
+ bool containsContinueInstruction() const;
+ bool containsLeaveInstruction() const;
+ bool usesSampler(int i) const;
+
+ struct Semantic
+ {
+ Semantic(unsigned char usage = 0xFF, unsigned char index = 0xFF) : usage(usage), index(index), centroid(false)
+ {
+ }
+
+ bool operator==(const Semantic &semantic) const
+ {
+ return usage == semantic.usage && index == semantic.index;
+ }
+
+ bool active() const
+ {
+ return usage != 0xFF;
+ }
+
+ unsigned char usage;
+ unsigned char index;
+ bool centroid;
+ };
+
+ virtual void analyze() = 0;
+
+ // FIXME: Private
+ unsigned int dirtyConstantsF;
+ unsigned int dirtyConstantsI;
+ unsigned int dirtyConstantsB;
+
+ bool dynamicallyIndexedTemporaries;
+ bool dynamicallyIndexedInput;
+ bool dynamicallyIndexedOutput;
+
+ protected:
+ void parse(const unsigned long *token);
+
+ void analyzeDirtyConstants();
+ void analyzeDynamicBranching();
+ void analyzeSamplers();
+ void analyzeCallSites();
+ void analyzeDynamicIndexing();
+ void markFunctionAnalysis(int functionLabel, Analysis flag);
+
+ ShaderType shaderType;
+
+ union
+ {
+ unsigned short version;
+
+ struct
+ {
+ unsigned char minorVersion;
+ unsigned char majorVersion;
+ };
+ };
+
+ std::vector<Instruction*> instruction;
+
+ unsigned short usedSamplers; // Bit flags
+
+ private:
+ const int serialID;
+ static volatile int serialCounter;
+
+ bool dynamicBranching;
+ bool containsBreak;
+ bool containsContinue;
+ bool containsLeave;
+ };
+}
+
+#endif // sw_Shader_hpp
diff --git a/src/Shader/ShaderCore.cpp b/src/Shader/ShaderCore.cpp
index 26facac..1fc2b2a 100644
--- a/src/Shader/ShaderCore.cpp
+++ b/src/Shader/ShaderCore.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -11,11 +11,459 @@
#include "ShaderCore.hpp"
-#include "Debug.hpp"
+#include "Renderer/Renderer.hpp"
+#include "Common/Debug.hpp"
namespace sw
{
- void ShaderCore::mov(Color4f &dst, Color4f &src, bool floorToInteger)
+ extern TranscendentalPrecision logPrecision;
+ extern TranscendentalPrecision expPrecision;
+ extern TranscendentalPrecision rcpPrecision;
+ extern TranscendentalPrecision rsqPrecision;
+
+ Vector4i::Vector4i()
+ {
+ }
+
+ Vector4i::Vector4i(unsigned short x, unsigned short y, unsigned short z, unsigned short w)
+ {
+ this->x = Short4(x);
+ this->y = Short4(y);
+ this->z = Short4(z);
+ this->w = Short4(w);
+ }
+
+ Vector4i::Vector4i(const Vector4i &rhs)
+ {
+ x = rhs.x;
+ y = rhs.y;
+ z = rhs.z;
+ w = rhs.w;
+ }
+
+ Vector4i &Vector4i::operator=(const Vector4i &rhs)
+ {
+ x = rhs.x;
+ y = rhs.y;
+ z = rhs.z;
+ w = rhs.w;
+
+ return *this;
+ }
+
+ Short4 &Vector4i::operator[](int i)
+ {
+ switch(i)
+ {
+ case 0: return x;
+ case 1: return y;
+ case 2: return z;
+ case 3: return w;
+ }
+
+ return x;
+ }
+
+ Vector4f::Vector4f()
+ {
+ }
+
+ Vector4f::Vector4f(float x, float y, float z, float w)
+ {
+ this->x = Float4(x);
+ this->y = Float4(y);
+ this->z = Float4(z);
+ this->w = Float4(w);
+ }
+
+ Vector4f::Vector4f(const Vector4f &rhs)
+ {
+ x = rhs.x;
+ y = rhs.y;
+ z = rhs.z;
+ w = rhs.w;
+ }
+
+ Vector4f &Vector4f::operator=(const Vector4f &rhs)
+ {
+ x = rhs.x;
+ y = rhs.y;
+ z = rhs.z;
+ w = rhs.w;
+
+ return *this;
+ }
+
+ Float4 &Vector4f::operator[](int i)
+ {
+ switch(i)
+ {
+ case 0: return x;
+ case 1: return y;
+ case 2: return z;
+ case 3: return w;
+ }
+
+ return x;
+ }
+
+ Float4 exponential2(RValue<Float4> x, bool pp)
+ {
+ Float4 x0;
+ Float4 x1;
+ Int4 x2;
+
+ x0 = x;
+
+ x0 = Min(x0, As<Float4>(Int4(0x43010000))); // 129.00000e+0f
+ x0 = Max(x0, As<Float4>(Int4(0xC2FDFFFF))); // -126.99999e+0f
+ x1 = x0;
+ x1 -= Float4(0.5f);
+ x2 = RoundInt(x1);
+ x1 = Float4(x2);
+ x2 += Int4(0x0000007F); // 127
+ x2 = x2 << 23;
+ x0 -= x1;
+ x1 = As<Float4>(Int4(0x3AF61905)); // 1.8775767e-3f
+ x1 *= x0;
+ x1 += As<Float4>(Int4(0x3C134806)); // 8.9893397e-3f
+ x1 *= x0;
+ x1 += As<Float4>(Int4(0x3D64AA23)); // 5.5826318e-2f
+ x1 *= x0;
+ x1 += As<Float4>(Int4(0x3E75EAD4)); // 2.4015361e-1f
+ x1 *= x0;
+ x1 += As<Float4>(Int4(0x3F31727B)); // 6.9315308e-1f
+ x1 *= x0;
+ x1 += As<Float4>(Int4(0x3F7FFFFF)); // 9.9999994e-1f
+ x1 *= As<Float4>(x2);
+
+ return x1;
+ }
+
+ Float4 logarithm2(RValue<Float4> x, bool absolute, bool pp)
+ {
+ Float4 x0;
+ Float4 x1;
+ Float4 x2;
+ Float4 x3;
+
+ x0 = x;
+
+ x1 = As<Float4>(As<Int4>(x0) & Int4(0x7F800000));
+ x1 = As<Float4>(As<UInt4>(x1) >> 8);
+ x1 = As<Float4>(As<Int4>(x1) | As<Int4>(Float4(1.0f)));
+ x1 = (x1 - Float4(1.4960938f)) * Float4(256.0f); // FIXME: (x1 - 1.4960938f) * 256.0f;
+ x0 = As<Float4>((As<Int4>(x0) & Int4(0x007FFFFF)) | As<Int4>(Float4(1.0f)));
+
+ x2 = (Float4(9.5428179e-2f) * x0 + Float4(4.7779095e-1f)) * x0 + Float4(1.9782813e-1f);
+ x3 = ((Float4(1.6618466e-2f) * x0 + Float4(2.0350508e-1f)) * x0 + Float4(2.7382900e-1f)) * x0 + Float4(4.0496687e-2f);
+ x2 /= x3;
+
+ x1 += (x0 - Float4(1.0f)) * x2;
+
+ return x1;
+ }
+
+ Float4 exponential(RValue<Float4> x, bool pp)
+ {
+ // FIXME: Propagate the constant
+ return exponential2(Float4(1.44269541f) * x, pp); // 1/ln(2)
+ }
+
+ Float4 logarithm(RValue<Float4> x, bool absolute, bool pp)
+ {
+ // FIXME: Propagate the constant
+ return Float4(6.93147181e-1f) * logarithm2(x, absolute, pp); // ln(2)
+ }
+
+ Float4 power(RValue<Float4> x, RValue<Float4> y, bool pp)
+ {
+ Float4 log = logarithm2(x, true, pp);
+ log *= y;
+ return exponential2(log, pp);
+ }
+
+ Float4 reciprocal(RValue<Float4> x, bool pp, bool finite)
+ {
+ Float4 rcp;
+
+ if(!pp && rcpPrecision >= WHQL)
+ {
+ rcp = Float4(1.0f) / x;
+ }
+ else
+ {
+ rcp = Rcp_pp(x);
+
+ if(!pp)
+ {
+ rcp = (rcp + rcp) - (x * rcp * rcp);
+ }
+ }
+
+ if(finite)
+ {
+ int big = 0x7F7FFFFF;
+ rcp = Min(rcp, Float4((float&)big));
+ }
+
+ return rcp;
+ }
+
+ Float4 reciprocalSquareRoot(RValue<Float4> x, bool absolute, bool pp)
+ {
+ Float4 abs = x;
+
+ if(absolute)
+ {
+ abs = Abs(abs);
+ }
+
+ Float4 rsq;
+
+ if(!pp && rsqPrecision >= IEEE)
+ {
+ rsq = Float4(1.0f) / Sqrt(abs);
+ }
+ else
+ {
+ rsq = RcpSqrt_pp(abs);
+
+ if(!pp)
+ {
+ rsq = rsq * (Float4(3.0f) - rsq * rsq * abs) * Float4(0.5f);
+ }
+ }
+
+ int big = 0x7F7FFFFF;
+ rsq = Min(rsq, Float4((float&)big));
+
+ return rsq;
+ }
+
+ Float4 modulo(RValue<Float4> x, RValue<Float4> y)
+ {
+ return x - y * Floor(x / y);
+ }
+
+ Float4 sine_pi(RValue<Float4> x, bool pp)
+ {
+ const Float4 A = Float4(-4.05284734e-1f); // -4/pi^2
+ const Float4 B = Float4(1.27323954e+0f); // 4/pi
+ const Float4 C = Float4(7.75160950e-1f);
+ const Float4 D = Float4(2.24839049e-1f);
+
+ // Parabola approximating sine
+ Float4 sin = x * (Abs(x) * A + B);
+
+ // Improve precision from 0.06 to 0.001
+ if(true)
+ {
+ sin = sin * (Abs(sin) * D + C);
+ }
+
+ return sin;
+ }
+
+ Float4 cosine_pi(RValue<Float4> x, bool pp)
+ {
+ // cos(x) = sin(x + pi/2)
+ Float4 y = x + Float4(1.57079632e+0f);
+
+ // Wrap around
+ y -= As<Float4>(CmpNLT(y, Float4(3.14159265e+0f)) & As<Int4>(Float4(6.28318530e+0f)));
+
+ return sine_pi(y, pp);
+ }
+
+ Float4 sine(RValue<Float4> x, bool pp)
+ {
+ // Reduce to [-0.5, 0.5] range
+ Float4 y = x * Float4(1.59154943e-1f); // 1/2pi
+ y = y - Round(y);
+
+ const Float4 A = Float4(-16.0f);
+ const Float4 B = Float4(8.0f);
+ const Float4 C = Float4(7.75160950e-1f);
+ const Float4 D = Float4(2.24839049e-1f);
+
+ // Parabola approximating sine
+ Float4 sin = y * (Abs(y) * A + B);
+
+ // Improve precision from 0.06 to 0.001
+ if(true)
+ {
+ sin = sin * (Abs(sin) * D + C);
+ }
+
+ return sin;
+ }
+
+ Float4 cosine(RValue<Float4> x, bool pp)
+ {
+ // cos(x) = sin(x + pi/2)
+ Float4 y = x + Float4(1.57079632e+0f);
+ return sine(y, pp);
+ }
+
+ Float4 tangent(RValue<Float4> x, bool pp)
+ {
+ return sine(x, pp) / cosine(x, pp);
+ }
+
+ Float4 arccos(RValue<Float4> x, bool pp)
+ {
+ // pi/2 - arcsin(x)
+ return Float4(1.57079632e+0f) - arcsin(x);
+ }
+
+ Float4 arcsin(RValue<Float4> x, bool pp)
+ {
+ // x*(pi/2-sqrt(1-x*x)*pi/5)
+ return x * (Float4(1.57079632e+0f) - Sqrt(Float4(1.0f) - x*x) * Float4(6.28318531e-1f));
+ }
+
+ Float4 arctan(RValue<Float4> x, bool pp)
+ {
+ Int4 O = CmpNLT(Abs(x), Float4(1.0f));
+ Float4 y = As<Float4>(O & As<Int4>(Float4(1.0f) / x) | ~O & As<Int4>(x)); // FIXME: Vector select
+
+ // Approximation of atan in [-1..1]
+ Float4 theta = y * (Float4(-0.27f) * Abs(y) + Float4(1.05539816f));
+
+ // +/-pi/2 depending on sign of x
+ Float4 sgnPi_2 = As<Float4>(As<Int4>(Float4(1.57079632e+0f)) ^ (As<Int4>(x) & Int4(0x80000000)));
+
+ theta = As<Float4>(O & As<Int4>(sgnPi_2 - theta) | ~O & As<Int4>(theta)); // FIXME: Vector select
+
+ return theta;
+ }
+
+ Float4 arctan(RValue<Float4> y, RValue<Float4> x, bool pp)
+ {
+ // Rotate to upper semicircle when in lower semicircle
+ Int4 S = CmpLT(y, Float4(0.0f));
+ Float4 theta = As<Float4>(S & As<Int4>(Float4(-3.14159265e+0f))); // -pi
+ Float4 x0 = As<Float4>((As<Int4>(y) & Int4(0x80000000)) ^ As<Int4>(x));
+ Float4 y0 = Abs(y);
+
+ // Rotate to right quadrant when in left quadrant
+ Int4 Q = CmpLT(x0, Float4(0.0f));
+ theta += As<Float4>(Q & As<Int4>(Float4(1.57079632e+0f))); // pi/2
+ Float4 x1 = As<Float4>(Q & As<Int4>(y0) | ~Q & As<Int4>(x0)); // FIXME: Vector select
+ Float4 y1 = As<Float4>(Q & As<Int4>(-x0) | ~Q & As<Int4>(y0)); // FIXME: Vector select
+
+ // Rotate to first octant when in second octant
+ Int4 O = CmpNLT(y1, x1);
+ theta += As<Float4>(O & As<Int4>(Float4(7.85398163e-1f))); // pi/4
+ Float4 x2 = As<Float4>(O & As<Int4>(Float4(7.07106781e-1f) * x1 + Float4(7.07106781e-1f) * y1) | ~O & As<Int4>(x1)); // sqrt(2)/2 // FIXME: Vector select
+ Float4 y2 = As<Float4>(O & As<Int4>(Float4(7.07106781e-1f) * y1 - Float4(7.07106781e-1f) * x1) | ~O & As<Int4>(y1)); // FIXME: Vector select
+
+ // Approximation of atan in [0..1]
+ Float4 y_x = y2 / x2;
+ theta += y_x * (Float4(-0.27f) * y_x + Float4(1.05539816f));
+
+ return theta;
+ }
+
+ Float4 dot2(Vector4f &v0, Vector4f &v1)
+ {
+ return v0.x * v1.x + v0.y * v1.y;
+ }
+
+ Float4 dot3(Vector4f &v0, Vector4f &v1)
+ {
+ return v0.x * v1.x + v0.y * v1.y + v0.z * v1.z;
+ }
+
+ Float4 dot4(Vector4f &v0, Vector4f &v1)
+ {
+ return v0.x * v1.x + v0.y * v1.y + v0.z * v1.z + v0.w * v1.w;
+ }
+
+ void transpose4x4(Short4 &row0, Short4 &row1, Short4 &row2, Short4 &row3)
+ {
+ Int2 tmp0 = UnpackHigh(row0, row1);
+ Int2 tmp1 = UnpackHigh(row2, row3);
+ Int2 tmp2 = UnpackLow(row0, row1);
+ Int2 tmp3 = UnpackLow(row2, row3);
+
+ row0 = As<Short4>(UnpackLow(tmp2, tmp3));
+ row1 = As<Short4>(UnpackHigh(tmp2, tmp3));
+ row2 = As<Short4>(UnpackLow(tmp0, tmp1));
+ row3 = As<Short4>(UnpackHigh(tmp0, tmp1));
+ }
+
+ void transpose4x4(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3)
+ {
+ Float4 tmp0 = UnpackLow(row0, row1);
+ Float4 tmp1 = UnpackLow(row2, row3);
+ Float4 tmp2 = UnpackHigh(row0, row1);
+ Float4 tmp3 = UnpackHigh(row2, row3);
+
+ row0 = Float4(tmp0.xy, tmp1.xy);
+ row1 = Float4(tmp0.zw, tmp1.zw);
+ row2 = Float4(tmp2.xy, tmp3.xy);
+ row3 = Float4(tmp2.zw, tmp3.zw);
+ }
+
+ void transpose4x3(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3)
+ {
+ Float4 tmp0 = UnpackLow(row0, row1);
+ Float4 tmp1 = UnpackLow(row2, row3);
+ Float4 tmp2 = UnpackHigh(row0, row1);
+ Float4 tmp3 = UnpackHigh(row2, row3);
+
+ row0 = Float4(tmp0.xy, tmp1.xy);
+ row1 = Float4(tmp0.zw, tmp1.zw);
+ row2 = Float4(tmp2.xy, tmp3.xy);
+ }
+
+ void transpose4x2(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3)
+ {
+ Float4 tmp0 = UnpackLow(row0, row1);
+ Float4 tmp1 = UnpackLow(row2, row3);
+
+ row0 = Float4(tmp0.xy, tmp1.xy);
+ row1 = Float4(tmp0.zw, tmp1.zw);
+ }
+
+ void transpose4x1(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3)
+ {
+ Float4 tmp0 = UnpackLow(row0, row1);
+ Float4 tmp1 = UnpackLow(row2, row3);
+
+ row0 = Float4(tmp0.xy, tmp1.xy);
+ }
+
+ void transpose2x4(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3)
+ {
+ row0 = UnpackLow(row0, row1);
+ row1 = Float4(row0.zw, row1.zw);
+ row2 = UnpackHigh(row0, row1);
+ row3 = Float4(row2.zw, row3.zw);
+ }
+
+ void transpose2x4h(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3)
+ {
+ row0 = UnpackLow(row2, row3);
+ row1 = Float4(row0.zw, row1.zw);
+ row2 = UnpackHigh(row2, row3);
+ row3 = Float4(row2.zw, row3.zw);
+ }
+
+ void transpose4xN(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3, int N)
+ {
+ switch(N)
+ {
+ case 1: transpose4x1(row0, row1, row2, row3); break;
+ case 2: transpose4x2(row0, row1, row2, row3); break;
+ case 3: transpose4x3(row0, row1, row2, row3); break;
+ case 4: transpose4x4(row0, row1, row2, row3); break;
+ }
+ }
+
+ void ShaderCore::mov(Vector4f &dst, Vector4f &src, bool floorToInteger)
{
if(floorToInteger)
{
@@ -27,7 +475,23 @@
}
}
- void ShaderCore::add(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::f2b(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = As<Float4>(CmpNEQ(src.x, Float4(0.0f)));
+ dst.y = As<Float4>(CmpNEQ(src.y, Float4(0.0f)));
+ dst.z = As<Float4>(CmpNEQ(src.z, Float4(0.0f)));
+ dst.w = As<Float4>(CmpNEQ(src.w, Float4(0.0f)));
+ }
+
+ void ShaderCore::b2f(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = As<Float4>(As<Int4>(src.x) & As<Int4>(Float4(1.0f)));
+ dst.y = As<Float4>(As<Int4>(src.y) & As<Int4>(Float4(1.0f)));
+ dst.z = As<Float4>(As<Int4>(src.z) & As<Int4>(Float4(1.0f)));
+ dst.w = As<Float4>(As<Int4>(src.w) & As<Int4>(Float4(1.0f)));
+ }
+
+ void ShaderCore::add(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
dst.x = src0.x + src1.x;
dst.y = src0.y + src1.y;
@@ -35,7 +499,7 @@
dst.w = src0.w + src1.w;
}
- void ShaderCore::sub(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::sub(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
dst.x = src0.x - src1.x;
dst.y = src0.y - src1.y;
@@ -43,7 +507,7 @@
dst.w = src0.w - src1.w;
}
- void ShaderCore::mad(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2)
+ void ShaderCore::mad(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2)
{
dst.x = src0.x * src1.x + src2.x;
dst.y = src0.y * src1.y + src2.y;
@@ -51,7 +515,7 @@
dst.w = src0.w * src1.w + src2.w;
}
- void ShaderCore::mul(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::mul(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
dst.x = src0.x * src1.x;
dst.y = src0.y * src1.y;
@@ -59,7 +523,7 @@
dst.w = src0.w * src1.w;
}
- void ShaderCore::rcp(Color4f &dst, Color4f &src, bool pp)
+ void ShaderCore::rcpx(Vector4f &dst, Vector4f &src, bool pp)
{
Float4 rcp = reciprocal(src.x, pp, true);
@@ -69,17 +533,126 @@
dst.w = rcp;
}
- void ShaderCore::rsq(Color4f &dst, Color4f &src, bool pp)
+ void ShaderCore::div(Vector4f &dst, Vector4f &src0, Vector4f &src1)
+ {
+ dst.x = src0.x / src1.x;
+ dst.y = src0.y / src1.y;
+ dst.z = src0.z / src1.z;
+ dst.w = src0.w / src1.w;
+ }
+
+ void ShaderCore::mod(Vector4f &dst, Vector4f &src0, Vector4f &src1)
+ {
+ dst.x = modulo(src0.x, src1.x);
+ dst.y = modulo(src0.y, src1.y);
+ dst.z = modulo(src0.z, src1.z);
+ dst.w = modulo(src0.w, src1.w);
+ }
+
+ void ShaderCore::rsqx(Vector4f &dst, Vector4f &src, bool pp)
{
Float4 rsq = reciprocalSquareRoot(src.x, true, pp);
- dst.r = rsq;
- dst.g = rsq;
- dst.b = rsq;
- dst.a = rsq;
+ dst.x = rsq;
+ dst.y = rsq;
+ dst.z = rsq;
+ dst.w = rsq;
}
- void ShaderCore::dp3(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::sqrt(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = Sqrt(src.x);
+ dst.y = Sqrt(src.y);
+ dst.z = Sqrt(src.z);
+ dst.w = Sqrt(src.w);
+ }
+
+ void ShaderCore::rsq(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = reciprocalSquareRoot(src.x, false, pp);
+ dst.y = reciprocalSquareRoot(src.y, false, pp);
+ dst.z = reciprocalSquareRoot(src.z, false, pp);
+ dst.w = reciprocalSquareRoot(src.w, false, pp);
+ }
+
+ void ShaderCore::len2(Float4 &dst, Vector4f &src, bool pp)
+ {
+ dst = Sqrt(dot2(src, src));
+ }
+
+ void ShaderCore::len3(Float4 &dst, Vector4f &src, bool pp)
+ {
+ dst = Sqrt(dot3(src, src));
+ }
+
+ void ShaderCore::len4(Float4 &dst, Vector4f &src, bool pp)
+ {
+ dst = Sqrt(dot4(src, src));
+ }
+
+ void ShaderCore::dist1(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp)
+ {
+ dst = Abs(src0.x - src1.x);
+ }
+
+ void ShaderCore::dist2(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp)
+ {
+ Float4 dx = src0.x - src1.x;
+ Float4 dy = src0.y - src1.y;
+ Float4 dot2 = dx * dx + dy * dy;
+ dst = Sqrt(dot2);
+ }
+
+ void ShaderCore::dist3(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp)
+ {
+ Float4 dx = src0.x - src1.x;
+ Float4 dy = src0.y - src1.y;
+ Float4 dz = src0.z - src1.z;
+ Float4 dot3 = dx * dx + dy * dy + dz * dz;
+ dst = Sqrt(dot3);
+ }
+
+ void ShaderCore::dist4(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp)
+ {
+ Float4 dx = src0.x - src1.x;
+ Float4 dy = src0.y - src1.y;
+ Float4 dz = src0.z - src1.z;
+ Float4 dw = src0.w - src1.w;
+ Float4 dot4 = dx * dx + dy * dy + dz * dz + dw * dw;
+ dst = Sqrt(dot4);
+ }
+
+ void ShaderCore::dp1(Vector4f &dst, Vector4f &src0, Vector4f &src1)
+ {
+ Float4 t = src0.x * src1.x;
+
+ dst.x = t;
+ dst.y = t;
+ dst.z = t;
+ dst.w = t;
+ }
+
+ void ShaderCore::dp2(Vector4f &dst, Vector4f &src0, Vector4f &src1)
+ {
+ Float4 t = dot2(src0, src1);
+
+ dst.x = t;
+ dst.y = t;
+ dst.z = t;
+ dst.w = t;
+ }
+
+ void ShaderCore::dp2add(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2)
+ {
+ Float4 t = dot2(src0, src1) + src2.x;
+
+ dst.x = t;
+ dst.y = t;
+ dst.z = t;
+ dst.w = t;
+ }
+
+ void ShaderCore::dp3(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
Float4 dot = dot3(src0, src1);
@@ -89,7 +662,7 @@
dst.w = dot;
}
- void ShaderCore::dp4(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::dp4(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
Float4 dot = dot4(src0, src1);
@@ -99,7 +672,7 @@
dst.w = dot;
}
- void ShaderCore::min(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::min(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
dst.x = Min(src0.x, src1.x);
dst.y = Min(src0.y, src1.y);
@@ -107,7 +680,7 @@
dst.w = Min(src0.w, src1.w);
}
- void ShaderCore::max(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::max(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
dst.x = Max(src0.x, src1.x);
dst.y = Max(src0.y, src1.y);
@@ -115,39 +688,25 @@
dst.w = Max(src0.w, src1.w);
}
- void ShaderCore::slt(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::slt(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
- Int4 xMask = As<Int4>(CmpLT(src0.x, src1.x));
- Int4 yMask = As<Int4>(CmpLT(src0.y, src1.y));
- Int4 zMask = As<Int4>(CmpLT(src0.z, src1.z));
- Int4 wMask = As<Int4>(CmpLT(src0.w, src1.w));
-
- Int4 iOne = As<Int4>(Float4(1, 1, 1, 1));
-
- dst.x = As<Float4>(xMask & iOne);
- dst.y = As<Float4>(yMask & iOne);
- dst.z = As<Float4>(zMask & iOne);
- dst.w = As<Float4>(wMask & iOne);
+ dst.x = As<Float4>(As<Int4>(CmpLT(src0.x, src1.x)) & As<Int4>(Float4(1.0f)));
+ dst.y = As<Float4>(As<Int4>(CmpLT(src0.y, src1.y)) & As<Int4>(Float4(1.0f)));
+ dst.z = As<Float4>(As<Int4>(CmpLT(src0.z, src1.z)) & As<Int4>(Float4(1.0f)));
+ dst.w = As<Float4>(As<Int4>(CmpLT(src0.w, src1.w)) & As<Int4>(Float4(1.0f)));
}
- void ShaderCore::sge(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::step(Vector4f &dst, Vector4f &edge, Vector4f &x)
{
- Int4 xMask = As<Int4>(CmpNLT(src0.x, src1.x));
- Int4 yMask = As<Int4>(CmpNLT(src0.y, src1.y));
- Int4 zMask = As<Int4>(CmpNLT(src0.z, src1.z));
- Int4 wMask = As<Int4>(CmpNLT(src0.w, src1.w));
-
- Int4 iOne = As<Int4>(Float4(1, 1, 1, 1));
-
- dst.x = As<Float4>(xMask & iOne);
- dst.y = As<Float4>(yMask & iOne);
- dst.z = As<Float4>(zMask & iOne);
- dst.w = As<Float4>(wMask & iOne);
+ dst.x = As<Float4>(CmpNLT(x.x, edge.x) & As<Int4>(Float4(1.0f)));
+ dst.y = As<Float4>(CmpNLT(x.y, edge.y) & As<Int4>(Float4(1.0f)));
+ dst.z = As<Float4>(CmpNLT(x.z, edge.z) & As<Int4>(Float4(1.0f)));
+ dst.w = As<Float4>(CmpNLT(x.w, edge.w) & As<Int4>(Float4(1.0f)));
}
- void ShaderCore::exp(Color4f &dst, Color4f &src, bool pp)
+ void ShaderCore::exp2x(Vector4f &dst, Vector4f &src, bool pp)
{
- Float4 exp = exponential(src.x, pp);
+ Float4 exp = exponential2(src.x, pp);
dst.x = exp;
dst.y = exp;
@@ -155,9 +714,25 @@
dst.w = exp;
}
- void ShaderCore::log(Color4f &dst, Color4f &src, bool pp)
+ void ShaderCore::exp2(Vector4f &dst, Vector4f &src, bool pp)
{
- Float4 log = logarithm(src.x, true, pp);
+ dst.x = exponential2(src.x, pp);
+ dst.y = exponential2(src.y, pp);
+ dst.z = exponential2(src.z, pp);
+ dst.w = exponential2(src.w, pp);
+ }
+
+ void ShaderCore::exp(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = exponential(src.x, pp);
+ dst.y = exponential(src.y, pp);
+ dst.z = exponential(src.z, pp);
+ dst.w = exponential(src.w, pp);
+ }
+
+ void ShaderCore::log2x(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ Float4 log = logarithm2(src.x, true, pp);
dst.x = log;
dst.y = log;
@@ -165,33 +740,50 @@
dst.w = log;
}
- void ShaderCore::lit(Color4f &dst, Color4f &src)
+ void ShaderCore::log2(Vector4f &dst, Vector4f &src, bool pp)
{
- dst.x = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- dst.y = Max(src.x, Float4(0.0f, 0.0f, 0.0f, 0.0f));
+ dst.x = logarithm2(src.x, pp);
+ dst.y = logarithm2(src.y, pp);
+ dst.z = logarithm2(src.z, pp);
+ dst.w = logarithm2(src.w, pp);
+ }
+
+ void ShaderCore::log(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = logarithm(src.x, false, pp);
+ dst.y = logarithm(src.y, false, pp);
+ dst.z = logarithm(src.z, false, pp);
+ dst.w = logarithm(src.w, false, pp);
+ }
+
+ void ShaderCore::lit(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = Float4(1.0f);
+ dst.y = Max(src.x, Float4(0.0f));
Float4 pow;
pow = src.w;
- pow = Min(pow, Float4(127.9961f, 127.9961f, 127.9961f, 127.9961f));
- pow = Max(pow, Float4(-127.9961f, -127.9961f, -127.9961f, -127.9961f));
+ pow = Min(pow, Float4(127.9961f));
+ pow = Max(pow, Float4(-127.9961f));
dst.z = power(src.y, pow);
- dst.z = As<Float4>(As<Int4>(dst.z) & CmpNLT(src.x, Float4(0.0f, 0.0f, 0.0f, 0.0f)));
- dst.z = As<Float4>(As<Int4>(dst.z) & CmpNLT(src.y, Float4(0.0f, 0.0f, 0.0f, 0.0f)));
+ dst.z = As<Float4>(As<Int4>(dst.z) & CmpNLT(src.x, Float4(0.0f)));
+ dst.z = As<Float4>(As<Int4>(dst.z) & CmpNLT(src.y, Float4(0.0f)));
- dst.w = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ dst.w = Float4(1.0f);
}
- void ShaderCore::dst(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::att(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
+ // Computes attenuation factors (1, d, d^2, 1/d) assuming src0 = d^2, src1 = 1/d
dst.x = 1;
dst.y = src0.y * src1.y;
dst.z = src0.z;
dst.w = src1.w;
}
- void ShaderCore::lrp(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2)
+ void ShaderCore::lrp(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2)
{
dst.x = src0.x * (src1.x - src2.x) + src2.x;
dst.y = src0.y * (src1.y - src2.y) + src2.y;
@@ -199,15 +791,47 @@
dst.w = src0.w * (src1.w - src2.w) + src2.w;
}
- void ShaderCore::frc(Color4f &dst, Color4f &src)
+ void ShaderCore::smooth(Vector4f &dst, Vector4f &edge0, Vector4f &edge1, Vector4f &x)
{
- dst.x = Fraction(src.x);
- dst.y = Fraction(src.y);
- dst.z = Fraction(src.z);
- dst.w = Fraction(src.w);
+ Float4 tx = Min(Max((x.x - edge0.x) / (edge1.x - edge0.x), Float4(0.0f)), Float4(1.0f)); dst.x = tx * tx * (Float4(3.0f) - Float4(2.0f) * tx);
+ Float4 ty = Min(Max((x.y - edge0.y) / (edge1.y - edge0.y), Float4(0.0f)), Float4(1.0f)); dst.y = ty * ty * (Float4(3.0f) - Float4(2.0f) * ty);
+ Float4 tz = Min(Max((x.z - edge0.z) / (edge1.z - edge0.z), Float4(0.0f)), Float4(1.0f)); dst.z = tz * tz * (Float4(3.0f) - Float4(2.0f) * tz);
+ Float4 tw = Min(Max((x.w - edge0.w) / (edge1.w - edge0.w), Float4(0.0f)), Float4(1.0f)); dst.w = tw * tw * (Float4(3.0f) - Float4(2.0f) * tw);
}
- void ShaderCore::pow(Color4f &dst, Color4f &src0, Color4f &src1, bool pp)
+ void ShaderCore::frc(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = Frac(src.x);
+ dst.y = Frac(src.y);
+ dst.z = Frac(src.z);
+ dst.w = Frac(src.w);
+ }
+
+ void ShaderCore::trunc(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = Trunc(src.x);
+ dst.y = Trunc(src.y);
+ dst.z = Trunc(src.z);
+ dst.w = Trunc(src.w);
+ }
+
+ void ShaderCore::floor(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = Floor(src.x);
+ dst.y = Floor(src.y);
+ dst.z = Floor(src.z);
+ dst.w = Floor(src.w);
+ }
+
+ void ShaderCore::ceil(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = Ceil(src.x);
+ dst.y = Ceil(src.y);
+ dst.z = Ceil(src.z);
+ dst.w = Ceil(src.w);
+ }
+
+ void ShaderCore::powx(Vector4f &dst, Vector4f &src0, Vector4f &src1, bool pp)
{
Float4 pow = power(src0.x, src1.x, pp);
@@ -217,14 +841,136 @@
dst.w = pow;
}
- void ShaderCore::crs(Color4f &dst, Color4f &src0, Color4f &src1)
+ void ShaderCore::pow(Vector4f &dst, Vector4f &src0, Vector4f &src1, bool pp)
+ {
+ dst.x = power(src0.x, src1.x, pp);
+ dst.y = power(src0.y, src1.y, pp);
+ dst.z = power(src0.z, src1.z, pp);
+ dst.w = power(src0.w, src1.w, pp);
+ }
+
+ void ShaderCore::crs(Vector4f &dst, Vector4f &src0, Vector4f &src1)
{
dst.x = src0.y * src1.z - src0.z * src1.y;
dst.y = src0.z * src1.x - src0.x * src1.z;
dst.z = src0.x * src1.y - src0.y * src1.x;
}
- void ShaderCore::sgn(Color4f &dst, Color4f &src)
+ void ShaderCore::forward1(Vector4f &dst, Vector4f &N, Vector4f &I, Vector4f &Nref)
+ {
+ Int4 flip = CmpNLT(Nref.x * I.x, Float4(0.0f)) & Int4(0x80000000);
+
+ dst.x = As<Float4>(flip ^ As<Int4>(N.x));
+ }
+
+ void ShaderCore::forward2(Vector4f &dst, Vector4f &N, Vector4f &I, Vector4f &Nref)
+ {
+ Int4 flip = CmpNLT(dot2(Nref, I), Float4(0.0f)) & Int4(0x80000000);
+
+ dst.x = As<Float4>(flip ^ As<Int4>(N.x));
+ dst.y = As<Float4>(flip ^ As<Int4>(N.y));
+ }
+
+ void ShaderCore::forward3(Vector4f &dst, Vector4f &N, Vector4f &I, Vector4f &Nref)
+ {
+ Int4 flip = CmpNLT(dot3(Nref, I), Float4(0.0f)) & Int4(0x80000000);
+
+ dst.x = As<Float4>(flip ^ As<Int4>(N.x));
+ dst.y = As<Float4>(flip ^ As<Int4>(N.y));
+ dst.z = As<Float4>(flip ^ As<Int4>(N.z));
+ }
+
+ void ShaderCore::forward4(Vector4f &dst, Vector4f &N, Vector4f &I, Vector4f &Nref)
+ {
+ Int4 flip = CmpNLT(dot4(Nref, I), Float4(0.0f)) & Int4(0x80000000);
+
+ dst.x = As<Float4>(flip ^ As<Int4>(N.x));
+ dst.y = As<Float4>(flip ^ As<Int4>(N.y));
+ dst.z = As<Float4>(flip ^ As<Int4>(N.z));
+ dst.w = As<Float4>(flip ^ As<Int4>(N.w));
+ }
+
+ void ShaderCore::reflect1(Vector4f &dst, Vector4f &I, Vector4f &N)
+ {
+ Float4 d = N.x * I.x;
+
+ dst.x = I.x - Float4(2.0f) * d * N.x;
+ }
+
+ void ShaderCore::reflect2(Vector4f &dst, Vector4f &I, Vector4f &N)
+ {
+ Float4 d = dot2(N, I);
+
+ dst.x = I.x - Float4(2.0f) * d * N.x;
+ dst.y = I.y - Float4(2.0f) * d * N.y;
+ }
+
+ void ShaderCore::reflect3(Vector4f &dst, Vector4f &I, Vector4f &N)
+ {
+ Float4 d = dot3(N, I);
+
+ dst.x = I.x - Float4(2.0f) * d * N.x;
+ dst.y = I.y - Float4(2.0f) * d * N.y;
+ dst.z = I.z - Float4(2.0f) * d * N.z;
+ }
+
+ void ShaderCore::reflect4(Vector4f &dst, Vector4f &I, Vector4f &N)
+ {
+ Float4 d = dot4(N, I);
+
+ dst.x = I.x - Float4(2.0f) * d * N.x;
+ dst.y = I.y - Float4(2.0f) * d * N.y;
+ dst.z = I.z - Float4(2.0f) * d * N.z;
+ dst.w = I.w - Float4(2.0f) * d * N.w;
+ }
+
+ void ShaderCore::refract1(Vector4f &dst, Vector4f &I, Vector4f &N, Float4 &eta)
+ {
+ Float4 d = N.x * I.x;
+ Float4 k = Float4(1.0f) - eta * eta * (Float4(1.0f) - d * d);
+ Int4 pos = CmpNLT(k, Float4(0.0f));
+ Float4 t = (eta * d + Sqrt(k));
+
+ dst.x = As<Float4>(pos & As<Int4>(eta * I.x - t * N.x));
+ }
+
+ void ShaderCore::refract2(Vector4f &dst, Vector4f &I, Vector4f &N, Float4 &eta)
+ {
+ Float4 d = dot2(N, I);
+ Float4 k = Float4(1.0f) - eta * eta * (Float4(1.0f) - d * d);
+ Int4 pos = CmpNLT(k, Float4(0.0f));
+ Float4 t = (eta * d + Sqrt(k));
+
+ dst.x = As<Float4>(pos & As<Int4>(eta * I.x - t * N.x));
+ dst.y = As<Float4>(pos & As<Int4>(eta * I.y - t * N.y));
+ }
+
+ void ShaderCore::refract3(Vector4f &dst, Vector4f &I, Vector4f &N, Float4 &eta)
+ {
+ Float4 d = dot3(N, I);
+ Float4 k = Float4(1.0f) - eta * eta * (Float4(1.0f) - d * d);
+ Int4 pos = CmpNLT(k, Float4(0.0f));
+ Float4 t = (eta * d + Sqrt(k));
+
+ dst.x = As<Float4>(pos & As<Int4>(eta * I.x - t * N.x));
+ dst.y = As<Float4>(pos & As<Int4>(eta * I.y - t * N.y));
+ dst.z = As<Float4>(pos & As<Int4>(eta * I.z - t * N.z));
+ }
+
+ void ShaderCore::refract4(Vector4f &dst, Vector4f &I, Vector4f &N, Float4 &eta)
+ {
+ Float4 d = dot4(N, I);
+ Float4 k = Float4(1.0f) - eta * eta * (Float4(1.0f) - d * d);
+ Int4 pos = CmpNLT(k, Float4(0.0f));
+ Float4 t = (eta * d + Sqrt(k));
+
+ dst.x = As<Float4>(pos & As<Int4>(eta * I.x - t * N.x));
+ dst.y = As<Float4>(pos & As<Int4>(eta * I.y - t * N.y));
+ dst.z = As<Float4>(pos & As<Int4>(eta * I.z - t * N.z));
+ dst.w = As<Float4>(pos & As<Int4>(eta * I.w - t * N.w));
+ }
+
+ void ShaderCore::sgn(Vector4f &dst, Vector4f &src)
{
sgn(dst.x, src.x);
sgn(dst.y, src.y);
@@ -232,15 +978,26 @@
sgn(dst.w, src.w);
}
- void ShaderCore::abs(Color4f &dst, Color4f &src)
+ void ShaderCore::abs(Vector4f &dst, Vector4f &src)
{
dst.x = Abs(src.x);
dst.y = Abs(src.y);
dst.z = Abs(src.z);
dst.w = Abs(src.w);
}
+
+ void ShaderCore::nrm2(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ Float4 dot = dot2(src, src);
+ Float4 rsq = reciprocalSquareRoot(dot, false, pp);
- void ShaderCore::nrm(Color4f &dst, Color4f &src, bool pp)
+ dst.x = src.x * rsq;
+ dst.y = src.y * rsq;
+ dst.z = src.z * rsq;
+ dst.w = src.w * rsq;
+ }
+
+ void ShaderCore::nrm3(Vector4f &dst, Vector4f &src, bool pp)
{
Float4 dot = dot3(src, src);
Float4 rsq = reciprocalSquareRoot(dot, false, pp);
@@ -250,42 +1007,99 @@
dst.z = src.z * rsq;
dst.w = src.w * rsq;
}
-
- void ShaderCore::sincos(Color4f &dst, Color4f &src, bool pp)
+
+ void ShaderCore::nrm4(Vector4f &dst, Vector4f &src, bool pp)
{
- Float4 tmp0;
- Float4 tmp1;
+ Float4 dot = dot4(src, src);
+ Float4 rsq = reciprocalSquareRoot(dot, false, pp);
- tmp0 = src.x;
-
- // cos(x) = sin(x + pi/2)
- tmp0 += Float4(1.57079632e+0f);
- tmp1 = As<Float4>(CmpNLT(tmp0, Float4(3.14159265e+0f)) & As<Int4>(Float4(6.28318530e+0f)));
- tmp0 -= tmp1;
-
- dst.x = sine(tmp0, pp);
- dst.y = sine(src.x, pp);
+ dst.x = src.x * rsq;
+ dst.y = src.y * rsq;
+ dst.z = src.z * rsq;
+ dst.w = src.w * rsq;
+ }
+
+ void ShaderCore::sincos(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = cosine_pi(src.x, pp);
+ dst.y = sine_pi(src.x, pp);
}
- void ShaderCore::expp(Color4f &dst, Color4f &src, unsigned short version)
+ void ShaderCore::cos(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = cosine(src.x, pp);
+ dst.y = cosine(src.y, pp);
+ dst.z = cosine(src.z, pp);
+ dst.w = cosine(src.w, pp);
+ }
+
+ void ShaderCore::sin(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = sine(src.x, pp);
+ dst.y = sine(src.y, pp);
+ dst.z = sine(src.z, pp);
+ dst.w = sine(src.w, pp);
+ }
+
+ void ShaderCore::tan(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = tangent(src.x, pp);
+ dst.y = tangent(src.y, pp);
+ dst.z = tangent(src.z, pp);
+ dst.w = tangent(src.w, pp);
+ }
+
+ void ShaderCore::acos(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = arccos(src.x, pp);
+ dst.y = arccos(src.y, pp);
+ dst.z = arccos(src.z, pp);
+ dst.w = arccos(src.w, pp);
+ }
+
+ void ShaderCore::asin(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = arcsin(src.x, pp);
+ dst.y = arcsin(src.y, pp);
+ dst.z = arcsin(src.z, pp);
+ dst.w = arcsin(src.w, pp);
+ }
+
+ void ShaderCore::atan(Vector4f &dst, Vector4f &src, bool pp)
+ {
+ dst.x = arctan(src.x, pp);
+ dst.y = arctan(src.y, pp);
+ dst.z = arctan(src.z, pp);
+ dst.w = arctan(src.w, pp);
+ }
+
+ void ShaderCore::atan2(Vector4f &dst, Vector4f &src0, Vector4f &src1, bool pp)
+ {
+ dst.x = arctan(src0.x, src1.x, pp);
+ dst.y = arctan(src0.y, src1.y, pp);
+ dst.z = arctan(src0.z, src1.z, pp);
+ dst.w = arctan(src0.w, src1.w, pp);
+ }
+
+ void ShaderCore::expp(Vector4f &dst, Vector4f &src, unsigned short version)
{
if(version < 0x0200)
{
- Float4 frc = Fraction(src.x);
+ Float4 frc = Frac(src.x);
Float4 floor = src.x - frc;
- dst.x = exponential(floor, true);
+ dst.x = exponential2(floor, true);
dst.y = frc;
- dst.z = exponential(src.x, true);
- dst.w = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ dst.z = exponential2(src.x, true);
+ dst.w = Float4(1.0f);
}
else // Version >= 2.0
{
- exp(dst, src, true); // FIXME: 10-bit precision suffices
+ exp2x(dst, src, true); // FIXME: 10-bit precision suffices
}
}
- void ShaderCore::logp(Color4f &dst, Color4f &src, unsigned short version)
+ void ShaderCore::logp(Vector4f &dst, Vector4f &src, unsigned short version)
{
if(version < 0x0200)
{
@@ -298,92 +1112,109 @@
tmp1 = tmp0;
// X component
- r = As<Int4>(As<UInt4>(tmp0) >> 23) - Int4(127, 127, 127, 127);
+ r = As<Int4>(As<UInt4>(tmp0) >> 23) - Int4(127);
dst.x = Float4(r);
// Y component
dst.y = As<Float4>((As<Int4>(tmp1) & Int4(0x007FFFFF)) | As<Int4>(Float4(1.0f)));
// Z component
- dst.z = logarithm(src.x, true, true);
+ dst.z = logarithm2(src.x, true, true);
// W component
dst.w = 1.0f;
}
else
{
- log(dst, src, true);
+ log2x(dst, src, true);
}
}
- void ShaderCore::cmp(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2)
+ void ShaderCore::cmp0(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2)
{
- cmp(dst.x, src0.x, src1.x, src2.x);
- cmp(dst.y, src0.y, src1.y, src2.y);
- cmp(dst.z, src0.z, src1.z, src2.z);
- cmp(dst.w, src0.w, src1.w, src2.w);
+ cmp0(dst.x, src0.x, src1.x, src2.x);
+ cmp0(dst.y, src0.y, src1.y, src2.y);
+ cmp0(dst.z, src0.z, src1.z, src2.z);
+ cmp0(dst.w, src0.w, src1.w, src2.w);
}
-
- void ShaderCore::dp2add(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2)
- {
- Float4 t = src0.x * src1.x + src0.y * src1.y + src2.x;
- dst.x = t;
- dst.y = t;
- dst.z = t;
- dst.w = t;
+ void ShaderCore::select(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2)
+ {
+ select(dst.x, As<Int4>(src0.x), src1.x, src2.x);
+ select(dst.y, As<Int4>(src0.y), src1.y, src2.y);
+ select(dst.z, As<Int4>(src0.z), src1.z, src2.z);
+ select(dst.w, As<Int4>(src0.w), src1.w, src2.w);
+ }
+
+ void ShaderCore::extract(Float4 &dst, Vector4f &src0, Float4 &src1)
+ {
+ select(dst, CmpEQ(src1, Float4(1.0f)), src0.y, src0.x);
+ select(dst, CmpEQ(src1, Float4(2.0f)), src0.z, dst);
+ select(dst, CmpEQ(src1, Float4(3.0f)), src0.w, dst);
+ }
+
+ void ShaderCore::insert(Vector4f &dst, Vector4f &src, Float4 &element, Float4 &index)
+ {
+ select(dst.x, CmpEQ(index, Float4(0.0f)), element, src.x);
+ select(dst.y, CmpEQ(index, Float4(1.0f)), element, src.y);
+ select(dst.z, CmpEQ(index, Float4(2.0f)), element, src.z);
+ select(dst.w, CmpEQ(index, Float4(3.0f)), element, src.w);
}
void ShaderCore::sgn(Float4 &dst, Float4 &src)
{
- Int4 neg = As<Int4>(CmpLT(src, Float4(0, 0, 0, 0))) & As<Int4>(Float4(-1, -1, -1, -1));
- Int4 pos = As<Int4>(CmpNLT(src, Float4(0, 0, 0, 0))) & As<Int4>(Float4(1, 1, 1, 1));
+ Int4 neg = As<Int4>(CmpLT(src, Float4(-0.0f))) & As<Int4>(Float4(-1.0f));
+ Int4 pos = As<Int4>(CmpNLE(src, Float4(+0.0f))) & As<Int4>(Float4(1.0f));
dst = As<Float4>(neg | pos);
}
- void ShaderCore::cmp(Float4 &dst, Float4 &src0, Float4 &src1, Float4 &src2)
+ void ShaderCore::cmp0(Float4 &dst, Float4 &src0, Float4 &src1, Float4 &src2)
{
- Int4 pos = CmpNLE(Float4(0.0f, 0.0f, 0.0f, 0.0f), src0);
- Int4 t0 = pos & As<Int4>(src2);
- Int4 t1 = ~pos & As<Int4>(src1);
- dst = As<Float4>(t0 | t1);
+ Int4 pos = CmpLE(Float4(0.0f), src0);
+ select(dst, pos, src1, src2);
}
- void ShaderCore::setp(Color4f &dst, Color4f &src0, Color4f &src1, Control control)
+ void ShaderCore::select(Float4 &dst, RValue<Int4> src0, Float4 &src1, Float4 &src2)
+ {
+ // FIXME: LLVM vector select
+ dst = As<Float4>(src0 & As<Int4>(src1) | ~src0 & As<Int4>(src2));
+ }
+
+ void ShaderCore::cmp(Vector4f &dst, Vector4f &src0, Vector4f &src1, Control control)
{
switch(control)
{
- case Op::CONTROL_GT:
+ case Shader::CONTROL_GT:
dst.x = As<Float4>(CmpNLE(src0.x, src1.x));
dst.y = As<Float4>(CmpNLE(src0.y, src1.y));
dst.z = As<Float4>(CmpNLE(src0.z, src1.z));
dst.w = As<Float4>(CmpNLE(src0.w, src1.w));
break;
- case Op::CONTROL_EQ:
+ case Shader::CONTROL_EQ:
dst.x = As<Float4>(CmpEQ(src0.x, src1.x));
dst.y = As<Float4>(CmpEQ(src0.y, src1.y));
dst.z = As<Float4>(CmpEQ(src0.z, src1.z));
dst.w = As<Float4>(CmpEQ(src0.w, src1.w));
break;
- case Op::CONTROL_GE:
+ case Shader::CONTROL_GE:
dst.x = As<Float4>(CmpNLT(src0.x, src1.x));
dst.y = As<Float4>(CmpNLT(src0.y, src1.y));
dst.z = As<Float4>(CmpNLT(src0.z, src1.z));
dst.w = As<Float4>(CmpNLT(src0.w, src1.w));
break;
- case Op::CONTROL_LT:
+ case Shader::CONTROL_LT:
dst.x = As<Float4>(CmpLT(src0.x, src1.x));
dst.y = As<Float4>(CmpLT(src0.y, src1.y));
dst.z = As<Float4>(CmpLT(src0.z, src1.z));
dst.w = As<Float4>(CmpLT(src0.w, src1.w));
break;
- case Op::CONTROL_NE:
+ case Shader::CONTROL_NE:
dst.x = As<Float4>(CmpNEQ(src0.x, src1.x));
dst.y = As<Float4>(CmpNEQ(src0.y, src1.y));
dst.z = As<Float4>(CmpNEQ(src0.z, src1.z));
dst.w = As<Float4>(CmpNEQ(src0.w, src1.w));
break;
- case Op::CONTROL_LE:
+ case Shader::CONTROL_LE:
dst.x = As<Float4>(CmpLE(src0.x, src1.x));
dst.y = As<Float4>(CmpLE(src0.y, src1.y));
dst.z = As<Float4>(CmpLE(src0.z, src1.z));
@@ -393,4 +1224,82 @@
ASSERT(false);
}
}
+
+ void ShaderCore::icmp(Vector4f &dst, Vector4f &src0, Vector4f &src1, Control control)
+ {
+ switch(control)
+ {
+ case Shader::CONTROL_GT:
+ dst.x = As<Float4>(CmpNLE(As<Int4>(src0.x), As<Int4>(src1.x)));
+ dst.y = As<Float4>(CmpNLE(As<Int4>(src0.y), As<Int4>(src1.y)));
+ dst.z = As<Float4>(CmpNLE(As<Int4>(src0.z), As<Int4>(src1.z)));
+ dst.w = As<Float4>(CmpNLE(As<Int4>(src0.w), As<Int4>(src1.w)));
+ break;
+ case Shader::CONTROL_EQ:
+ dst.x = As<Float4>(CmpEQ(As<Int4>(src0.x), As<Int4>(src1.x)));
+ dst.y = As<Float4>(CmpEQ(As<Int4>(src0.y), As<Int4>(src1.y)));
+ dst.z = As<Float4>(CmpEQ(As<Int4>(src0.z), As<Int4>(src1.z)));
+ dst.w = As<Float4>(CmpEQ(As<Int4>(src0.w), As<Int4>(src1.w)));
+ break;
+ case Shader::CONTROL_GE:
+ dst.x = As<Float4>(CmpNLT(As<Int4>(src0.x), As<Int4>(src1.x)));
+ dst.y = As<Float4>(CmpNLT(As<Int4>(src0.y), As<Int4>(src1.y)));
+ dst.z = As<Float4>(CmpNLT(As<Int4>(src0.z), As<Int4>(src1.z)));
+ dst.w = As<Float4>(CmpNLT(As<Int4>(src0.w), As<Int4>(src1.w)));
+ break;
+ case Shader::CONTROL_LT:
+ dst.x = As<Float4>(CmpLT(As<Int4>(src0.x), As<Int4>(src1.x)));
+ dst.y = As<Float4>(CmpLT(As<Int4>(src0.y), As<Int4>(src1.y)));
+ dst.z = As<Float4>(CmpLT(As<Int4>(src0.z), As<Int4>(src1.z)));
+ dst.w = As<Float4>(CmpLT(As<Int4>(src0.w), As<Int4>(src1.w)));
+ break;
+ case Shader::CONTROL_NE:
+ dst.x = As<Float4>(CmpNEQ(As<Int4>(src0.x), As<Int4>(src1.x)));
+ dst.y = As<Float4>(CmpNEQ(As<Int4>(src0.y), As<Int4>(src1.y)));
+ dst.z = As<Float4>(CmpNEQ(As<Int4>(src0.z), As<Int4>(src1.z)));
+ dst.w = As<Float4>(CmpNEQ(As<Int4>(src0.w), As<Int4>(src1.w)));
+ break;
+ case Shader::CONTROL_LE:
+ dst.x = As<Float4>(CmpLE(As<Int4>(src0.x), As<Int4>(src1.x)));
+ dst.y = As<Float4>(CmpLE(As<Int4>(src0.y), As<Int4>(src1.y)));
+ dst.z = As<Float4>(CmpLE(As<Int4>(src0.z), As<Int4>(src1.z)));
+ dst.w = As<Float4>(CmpLE(As<Int4>(src0.w), As<Int4>(src1.w)));
+ break;
+ default:
+ ASSERT(false);
+ }
+ }
+
+ void ShaderCore::all(Float4 &dst, Vector4f &src)
+ {
+ dst = As<Float4>(As<Int4>(src.x) & As<Int4>(src.y) & As<Int4>(src.z) & As<Int4>(src.w));
+ }
+
+ void ShaderCore::any(Float4 &dst, Vector4f &src)
+ {
+ dst = As<Float4>(As<Int4>(src.x) | As<Int4>(src.y) | As<Int4>(src.z) | As<Int4>(src.w));
+ }
+
+ void ShaderCore::not(Vector4f &dst, Vector4f &src)
+ {
+ dst.x = As<Float4>(As<Int4>(src.x) ^ Int4(0xFFFFFFFF));
+ dst.y = As<Float4>(As<Int4>(src.y) ^ Int4(0xFFFFFFFF));
+ dst.z = As<Float4>(As<Int4>(src.z) ^ Int4(0xFFFFFFFF));
+ dst.w = As<Float4>(As<Int4>(src.w) ^ Int4(0xFFFFFFFF));
+ }
+
+ void ShaderCore::or(Float4 &dst, Float4 &src0, Float4 &src1)
+ {
+ dst = As<Float4>(As<Int4>(src0) | As<Int4>(src1));
+ }
+
+ void ShaderCore::xor(Float4 &dst, Float4 &src0, Float4 &src1)
+ {
+ dst = As<Float4>(As<Int4>(src0) ^ As<Int4>(src1));
+ }
+
+ void ShaderCore::and(Float4 &dst, Float4 &src0, Float4 &src1)
+ {
+ dst = As<Float4>(As<Int4>(src0) & As<Int4>(src1));
+ }
}
diff --git a/src/Shader/ShaderCore.hpp b/src/Shader/ShaderCore.hpp
index 662e50a..6cb42f4 100644
--- a/src/Shader/ShaderCore.hpp
+++ b/src/Shader/ShaderCore.hpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -12,51 +12,284 @@
#ifndef sw_ShaderCore_hpp
#define sw_ShaderCore_hpp
+#include "Debug.hpp"
#include "Shader.hpp"
#include "Reactor/Reactor.hpp"
namespace sw
{
+ class Vector4i
+ {
+ public:
+ Vector4i();
+ Vector4i(unsigned short x, unsigned short y, unsigned short z, unsigned short w);
+ Vector4i(const Vector4i &rhs);
+
+ Short4 &operator[](int i);
+ Vector4i &operator=(const Vector4i &rhs);
+
+ Short4 x;
+ Short4 y;
+ Short4 z;
+ Short4 w;
+ };
+
+ class Vector4f
+ {
+ public:
+ Vector4f();
+ Vector4f(float x, float y, float z, float w);
+ Vector4f(const Vector4f &rhs);
+
+ Float4 &operator[](int i);
+ Vector4f &operator=(const Vector4f &rhs);
+
+ Float4 x;
+ Float4 y;
+ Float4 z;
+ Float4 w;
+ };
+
+ Float4 exponential2(RValue<Float4> x, bool pp = false);
+ Float4 logarithm2(RValue<Float4> x, bool abs, bool pp = false);
+ Float4 exponential(RValue<Float4> x, bool pp = false);
+ Float4 logarithm(RValue<Float4> x, bool abs, bool pp = false);
+ Float4 power(RValue<Float4> x, RValue<Float4> y, bool pp = false);
+ Float4 reciprocal(RValue<Float4> x, bool pp = false, bool finite = false);
+ Float4 reciprocalSquareRoot(RValue<Float4> x, bool abs, bool pp = false);
+ Float4 modulo(RValue<Float4> x, RValue<Float4> y);
+ Float4 sine_pi(RValue<Float4> x, bool pp = false); // limited to [-pi, pi] range
+ Float4 cosine_pi(RValue<Float4> x, bool pp = false); // limited to [-pi, pi] range
+ Float4 sine(RValue<Float4> x, bool pp = false);
+ Float4 cosine(RValue<Float4> x, bool pp = false);
+ Float4 tangent(RValue<Float4> x, bool pp = false);
+ Float4 arccos(RValue<Float4> x, bool pp = false);
+ Float4 arcsin(RValue<Float4> x, bool pp = false);
+ Float4 arctan(RValue<Float4> x, bool pp = false);
+ Float4 arctan(RValue<Float4> y, RValue<Float4> x, bool pp = false);
+
+ Float4 dot2(Vector4f &v0, Vector4f &v1);
+ Float4 dot3(Vector4f &v0, Vector4f &v1);
+ Float4 dot4(Vector4f &v0, Vector4f &v1);
+
+ void transpose4x4(Short4 &row0, Short4 &row1, Short4 &row2, Short4 &row3);
+ void transpose4x4(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3);
+ void transpose4x3(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3);
+ void transpose4x2(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3);
+ void transpose4x1(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3);
+ void transpose2x4(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3);
+ void transpose2x4h(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3);
+ void transpose4xN(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3, int N);
+
+ class Register
+ {
+ public:
+ Register(Reference<Float4> &x, Reference<Float4> &y, Reference<Float4> &z, Reference<Float4> &w) : x(x), y(y), z(z), w(w)
+ {
+ }
+
+ Reference<Float4> &operator[](int i)
+ {
+ switch(i)
+ {
+ default:
+ case 0: return x;
+ case 1: return y;
+ case 2: return z;
+ case 3: return w;
+ }
+ }
+
+ Register &operator=(const Register &rhs)
+ {
+ x = rhs.x;
+ y = rhs.y;
+ z = rhs.z;
+ w = rhs.w;
+
+ return *this;
+ }
+
+ Register &operator=(const Vector4f &rhs)
+ {
+ x = rhs.x;
+ y = rhs.y;
+ z = rhs.z;
+ w = rhs.w;
+
+ return *this;
+ }
+
+ operator Vector4f()
+ {
+ Vector4f v;
+
+ v.x = x;
+ v.y = y;
+ v.z = z;
+ v.w = w;
+
+ return v;
+ }
+
+ Reference<Float4> x;
+ Reference<Float4> y;
+ Reference<Float4> z;
+ Reference<Float4> w;
+ };
+
+ template<int S, bool D = false>
+ class RegisterArray
+ {
+ public:
+ RegisterArray(bool dynamic = D) : dynamic(dynamic)
+ {
+ if(dynamic)
+ {
+ x = new Array<Float4>(S);
+ y = new Array<Float4>(S);
+ z = new Array<Float4>(S);
+ w = new Array<Float4>(S);
+ }
+ else
+ {
+ x = new Array<Float4>[S];
+ y = new Array<Float4>[S];
+ z = new Array<Float4>[S];
+ w = new Array<Float4>[S];
+ }
+ }
+
+ ~RegisterArray()
+ {
+ delete[] x;
+ delete[] y;
+ delete[] z;
+ delete[] w;
+ }
+
+ Register operator[](int i)
+ {
+ if(dynamic)
+ {
+ return Register(x[0][i], y[0][i], z[0][i], w[0][i]);
+ }
+ else
+ {
+ return Register(x[i][0], y[i][0], z[i][0], w[i][0]);
+ }
+ }
+
+ Register operator[](RValue<Int> i)
+ {
+ ASSERT(dynamic);
+
+ return Register(x[0][i], y[0][i], z[0][i], w[0][i]);
+ }
+
+ private:
+ const bool dynamic;
+ Array<Float4> *x;
+ Array<Float4> *y;
+ Array<Float4> *z;
+ Array<Float4> *w;
+ };
+
class ShaderCore
{
- typedef Shader::Instruction::Operation::Control Control;
- typedef Shader::Instruction::Operation Op;
+ typedef Shader::Control Control;
public:
- void mov(Color4f &dst, Color4f &src, bool floorToInteger = false);
- void add(Color4f &dst, Color4f &src0, Color4f &src1);
- void sub(Color4f &dst, Color4f &src0, Color4f &src1);
- void mad(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2);
- void mul(Color4f &dst, Color4f &src0, Color4f &src1);
- void rcp(Color4f &dst, Color4f &src, bool pp = false);
- void rsq(Color4f &dst, Color4f &src, bool pp = false);
- void dp3(Color4f &dst, Color4f &src0, Color4f &src1);
- void dp4(Color4f &dst, Color4f &src0, Color4f &src1);
- void min(Color4f &dst, Color4f &src0, Color4f &src1);
- void max(Color4f &dst, Color4f &src0, Color4f &src1);
- void slt(Color4f &dst, Color4f &src0, Color4f &src1);
- void sge(Color4f &dst, Color4f &src0, Color4f &src1);
- void exp(Color4f &dst, Color4f &src, bool pp = false);
- void log(Color4f &dst, Color4f &src, bool pp = false);
- void lit(Color4f &dst, Color4f &src);
- void dst(Color4f &dst, Color4f &src0, Color4f &src1);
- void lrp(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2);
- void frc(Color4f &dst, Color4f &src);
- void pow(Color4f &dst, Color4f &src0, Color4f &src1, bool pp = false);
- void crs(Color4f &dst, Color4f &src0, Color4f &src1);
- void sgn(Color4f &dst, Color4f &src);
- void abs(Color4f &dst, Color4f &src);
- void nrm(Color4f &dst, Color4f &src, bool pp = false);
- void sincos(Color4f &dst, Color4f &src, bool pp = false);
- void expp(Color4f &dst, Color4f &src, unsigned short version);
- void logp(Color4f &dst, Color4f &src, unsigned short version);
- void cmp(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2);
- void dp2add(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2);
- void setp(Color4f &dst, Color4f &src0, Color4f &src1, Control control);
+ void mov(Vector4f &dst, Vector4f &src, bool floorToInteger = false);
+ void f2b(Vector4f &dst, Vector4f &src);
+ void b2f(Vector4f &dst, Vector4f &src);
+ void add(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void sub(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void mad(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void mul(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void rcpx(Vector4f &dst, Vector4f &src, bool pp = false);
+ void div(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void mod(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void rsqx(Vector4f &dst, Vector4f &src, bool pp = false);
+ void sqrt(Vector4f &dst, Vector4f &src, bool pp = false);
+ void rsq(Vector4f &dst, Vector4f &src, bool pp = false);
+ void len2(Float4 &dst, Vector4f &src, bool pp = false);
+ void len3(Float4 &dst, Vector4f &src, bool pp = false);
+ void len4(Float4 &dst, Vector4f &src, bool pp = false);
+ void dist1(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp = false);
+ void dist2(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp = false);
+ void dist3(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp = false);
+ void dist4(Float4 &dst, Vector4f &src0, Vector4f &src1, bool pp = false);
+ void dp1(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void dp2(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void dp2add(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void dp3(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void dp4(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void min(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void max(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void slt(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void step(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void exp2x(Vector4f &dst, Vector4f &src, bool pp = false);
+ void exp2(Vector4f &dst, Vector4f &src, bool pp = false);
+ void exp(Vector4f &dst, Vector4f &src, bool pp = false);
+ void log2x(Vector4f &dst, Vector4f &src, bool pp = false);
+ void log2(Vector4f &dst, Vector4f &src, bool pp = false);
+ void log(Vector4f &dst, Vector4f &src, bool pp = false);
+ void lit(Vector4f &dst, Vector4f &src);
+ void att(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void lrp(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void smooth(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void frc(Vector4f &dst, Vector4f &src);
+ void trunc(Vector4f &dst, Vector4f &src);
+ void floor(Vector4f &dst, Vector4f &src);
+ void ceil(Vector4f &dst, Vector4f &src);
+ void powx(Vector4f &dst, Vector4f &src0, Vector4f &src1, bool pp = false);
+ void pow(Vector4f &dst, Vector4f &src0, Vector4f &src1, bool pp = false);
+ void crs(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void forward1(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void forward2(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void forward3(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void forward4(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void reflect1(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void reflect2(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void reflect3(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void reflect4(Vector4f &dst, Vector4f &src0, Vector4f &src1);
+ void refract1(Vector4f &dst, Vector4f &src0, Vector4f &src1, Float4 &src2);
+ void refract2(Vector4f &dst, Vector4f &src0, Vector4f &src1, Float4 &src2);
+ void refract3(Vector4f &dst, Vector4f &src0, Vector4f &src1, Float4 &src2);
+ void refract4(Vector4f &dst, Vector4f &src0, Vector4f &src1, Float4 &src2);
+ void sgn(Vector4f &dst, Vector4f &src);
+ void abs(Vector4f &dst, Vector4f &src);
+ void nrm2(Vector4f &dst, Vector4f &src, bool pp = false);
+ void nrm3(Vector4f &dst, Vector4f &src, bool pp = false);
+ void nrm4(Vector4f &dst, Vector4f &src, bool pp = false);
+ void sincos(Vector4f &dst, Vector4f &src, bool pp = false);
+ void cos(Vector4f &dst, Vector4f &src, bool pp = false);
+ void sin(Vector4f &dst, Vector4f &src, bool pp = false);
+ void tan(Vector4f &dst, Vector4f &src, bool pp = false);
+ void acos(Vector4f &dst, Vector4f &src, bool pp = false);
+ void asin(Vector4f &dst, Vector4f &src, bool pp = false);
+ void atan(Vector4f &dst, Vector4f &src, bool pp = false);
+ void atan2(Vector4f &dst, Vector4f &src0, Vector4f &src1, bool pp = false);
+ void expp(Vector4f &dst, Vector4f &src, unsigned short version);
+ void logp(Vector4f &dst, Vector4f &src, unsigned short version);
+ void cmp0(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void cmp(Vector4f &dst, Vector4f &src0, Vector4f &src1, Control control);
+ void icmp(Vector4f &dst, Vector4f &src0, Vector4f &src1, Control control);
+ void select(Vector4f &dst, Vector4f &src0, Vector4f &src1, Vector4f &src2);
+ void extract(Float4 &dst, Vector4f &src0, Float4 &src1);
+ void insert(Vector4f &dst, Vector4f &src, Float4 &element, Float4 &index);
+ void all(Float4 &dst, Vector4f &src);
+ void any(Float4 &dst, Vector4f &src);
+ void not(Vector4f &dst, Vector4f &src);
+ void or(Float4 &dst, Float4 &src0, Float4 &src1);
+ void xor(Float4 &dst, Float4 &src0, Float4 &src1);
+ void and(Float4 &dst, Float4 &src0, Float4 &src1);
private:
void sgn(Float4 &dst, Float4 &src);
- void cmp(Float4 &dst, Float4 &src0, Float4 &src1, Float4 &src2);
+ void cmp0(Float4 &dst, Float4 &src0, Float4 &src1, Float4 &src2);
+ void select(Float4 &dst, RValue<Int4> src0, Float4 &src1, Float4 &src2);
};
}
diff --git a/src/Shader/VertexPipeline.cpp b/src/Shader/VertexPipeline.cpp
index 112b451..1e39ca7 100644
--- a/src/Shader/VertexPipeline.cpp
+++ b/src/Shader/VertexPipeline.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -24,7 +24,7 @@
namespace sw
{
- VertexPipeline::VertexPipeline(const VertexProcessor::State &state) : VertexRoutine(state)
+ VertexPipeline::VertexPipeline(const VertexProcessor::State &state) : VertexRoutine(state, 0)
{
}
@@ -32,9 +32,9 @@
{
}
- Color4f VertexPipeline::transformBlend(Registers &r, Color4f &src, Pointer<Byte> &matrix, bool homogeneous)
+ Vector4f VertexPipeline::transformBlend(Registers &r, Register &src, Pointer<Byte> &matrix, bool homogeneous)
{
- Color4f dst;
+ Vector4f dst;
if(state.vertexBlendMatrixCount == 0)
{
@@ -51,14 +51,15 @@
{
for(int i = 0; i < 4; i++)
{
+ Float4 B = r.v[BlendIndices].x;
UInt indices;
switch(i)
{
- case 0: indices = As<UInt>(Float(r.v[BlendIndices].x.x)); break;
- case 1: indices = As<UInt>(Float(r.v[BlendIndices].x.y)); break;
- case 2: indices = As<UInt>(Float(r.v[BlendIndices].x.z)); break;
- case 3: indices = As<UInt>(Float(r.v[BlendIndices].x.w)); break;
+ case 0: indices = As<UInt>(Float(B.x)); break;
+ case 1: indices = As<UInt>(Float(B.y)); break;
+ case 2: indices = As<UInt>(Float(B.z)); break;
+ case 3: indices = As<UInt>(Float(B.w)); break;
}
index0[i] = (indices & UInt(0x000000FF)) << UInt(6); // FIXME: (indices & 0x000000FF) << 6
@@ -100,13 +101,13 @@
{
weight1 = Float4(1.0f) - weight0;
- Color4f pos0;
- Color4f pos1;
+ Vector4f pos0;
+ Vector4f pos1;
pos0 = transform(src, matrix, index0, homogeneous);
pos1 = transform(src, matrix, index1, homogeneous);
- dst.x = pos0.x * weight0 + pos1.x * weight1; // FIXME: Color4f operators
+ dst.x = pos0.x * weight0 + pos1.x * weight1; // FIXME: Vector4f operators
dst.y = pos0.y * weight0 + pos1.y * weight1;
dst.z = pos0.z * weight0 + pos1.z * weight1;
dst.w = pos0.w * weight0 + pos1.w * weight1;
@@ -115,9 +116,9 @@
{
weight2 = Float4(1.0f) - (weight0 + weight1);
- Color4f pos0;
- Color4f pos1;
- Color4f pos2;
+ Vector4f pos0;
+ Vector4f pos1;
+ Vector4f pos2;
pos0 = transform(src, matrix, index0, homogeneous);
pos1 = transform(src, matrix, index1, homogeneous);
@@ -132,10 +133,10 @@
{
weight3 = Float4(1.0f) - (weight0 + weight1 + weight2);
- Color4f pos0;
- Color4f pos1;
- Color4f pos2;
- Color4f pos3;
+ Vector4f pos0;
+ Vector4f pos1;
+ Vector4f pos2;
+ Vector4f pos3;
pos0 = transform(src, matrix, index0, homogeneous);
pos1 = transform(src, matrix, index1, homogeneous);
@@ -154,8 +155,8 @@
void VertexPipeline::pipeline(Registers &r)
{
- Color4f position;
- Color4f normal;
+ Vector4f position;
+ Vector4f normal;
if(!state.preTransformed)
{
@@ -166,10 +167,10 @@
position = r.v[PositionT];
}
- r.ox[Pos] = position.x;
- r.oy[Pos] = position.y;
- r.oz[Pos] = position.z;
- r.ow[Pos] = position.w;
+ r.o[Pos].x = position.x;
+ r.o[Pos].y = position.y;
+ r.o[Pos].z = position.z;
+ r.o[Pos].w = position.w;
if(state.vertexNormalActive)
{
@@ -186,59 +187,59 @@
// FIXME: Don't process if not used at all
if(state.diffuseActive && state.input[Color0])
{
- Color4f diffuse = r.v[Color0];
+ Vector4f diffuse = r.v[Color0];
- r.ox[D0] = diffuse.x;
- r.oy[D0] = diffuse.y;
- r.oz[D0] = diffuse.z;
- r.ow[D0] = diffuse.w;
+ r.o[D0].x = diffuse.x;
+ r.o[D0].y = diffuse.y;
+ r.o[D0].z = diffuse.z;
+ r.o[D0].w = diffuse.w;
}
else
{
- r.ox[D0] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- r.oy[D0] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- r.oz[D0] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- r.ow[D0] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ r.o[D0].x = Float4(1.0f);
+ r.o[D0].y = Float4(1.0f);
+ r.o[D0].z = Float4(1.0f);
+ r.o[D0].w = Float4(1.0f);
}
// FIXME: Don't process if not used at all
if(state.specularActive && state.input[Color1])
{
- Color4f specular = r.v[Color1];
+ Vector4f specular = r.v[Color1];
- r.ox[D1] = specular.x;
- r.oy[D1] = specular.y;
- r.oz[D1] = specular.z;
- r.ow[D1] = specular.w;
+ r.o[D1].x = specular.x;
+ r.o[D1].y = specular.y;
+ r.o[D1].z = specular.z;
+ r.o[D1].w = specular.w;
}
else
{
- r.ox[D1] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.oy[D1] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.oz[D1] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.ow[D1] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ r.o[D1].x = Float4(0.0f);
+ r.o[D1].y = Float4(0.0f);
+ r.o[D1].z = Float4(0.0f);
+ r.o[D1].w = Float4(1.0f);
}
}
else
{
- Color4f diffuseSum;
+ Vector4f diffuseSum;
- r.ox[D0] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.oy[D0] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.oz[D0] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.ow[D0] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ r.o[D0].x = Float4(0.0f);
+ r.o[D0].y = Float4(0.0f);
+ r.o[D0].z = Float4(0.0f);
+ r.o[D0].w = Float4(0.0f);
- r.ox[D1] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.oy[D1] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.oz[D1] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.ow[D1] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ r.o[D1].x = Float4(0.0f);
+ r.o[D1].y = Float4(0.0f);
+ r.o[D1].z = Float4(0.0f);
+ r.o[D1].w = Float4(0.0f);
- diffuseSum.x = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- diffuseSum.y = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- diffuseSum.z = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- diffuseSum.w = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ diffuseSum.x = Float4(0.0f);
+ diffuseSum.y = Float4(0.0f);
+ diffuseSum.z = Float4(0.0f);
+ diffuseSum.w = Float4(0.0f);
- Color4f vertexPosition = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true);
+ Vector4f vertexPosition = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true);
for(int i = 0; i < 8; i++)
{
@@ -247,7 +248,7 @@
continue;
}
- Color4f L; // Light vector
+ Vector4f L; // Light vector
Float4 att; // Attenuation
// Attenuation
@@ -280,9 +281,9 @@
{
Float4 lightAmbient = *Pointer<Float4>(r.data + OFFSET(DrawData,ff.lightAmbient[i])); // FIXME: Unpack
- r.ox[D0] = r.ox[D0] + lightAmbient.x * att;
- r.oy[D0] = r.oy[D0] + lightAmbient.y * att;
- r.oz[D0] = r.oz[D0] + lightAmbient.z * att;
+ r.o[D0].x = r.o[D0].x + lightAmbient.x * att;
+ r.o[D0].y = r.o[D0].y + lightAmbient.y * att;
+ r.o[D0].z = r.o[D0].z + lightAmbient.z * att;
}
// Diffuse
@@ -291,10 +292,10 @@
Float4 dot;
dot = dot3(L, normal);
- dot = Max(dot, Float4(0.0f, 0.0f, 0.0f, 0.0f));
+ dot = Max(dot, Float4(0.0f));
dot *= att;
- Color4f diff;
+ Vector4f diff;
if(state.vertexDiffuseMaterialSourceActive == Context::MATERIAL)
{
@@ -323,15 +324,15 @@
// Specular
if(state.vertexSpecularActive)
{
- Color4f S;
- Color4f C; // Camera vector
+ Vector4f S;
+ Vector4f C; // Camera vector
Float4 pow;
pow = *Pointer<Float>(r.data + OFFSET(DrawData,ff.materialShininess));
- S.x = Float4(0.0f, 0.0f, 0.0f, 0.0f) - vertexPosition.x;
- S.y = Float4(0.0f, 0.0f, 0.0f, 0.0f) - vertexPosition.y;
- S.z = Float4(0.0f, 0.0f, 0.0f, 0.0f) - vertexPosition.z;
+ S.x = Float4(0.0f) - vertexPosition.x;
+ S.y = Float4(0.0f) - vertexPosition.y;
+ S.z = Float4(0.0f) - vertexPosition.z;
C = normalize(S);
S.x = L.x + C.x;
@@ -344,7 +345,7 @@
Float4 P = power(dot, pow);
P *= att;
- Color4f spec;
+ Vector4f spec;
if(state.vertexSpecularMaterialSourceActive == Context::MATERIAL)
{
@@ -374,93 +375,93 @@
spec.y *= P;
spec.z *= P;
- spec.x = Max(spec.x, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- spec.y = Max(spec.y, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- spec.z = Max(spec.z, Float4(0.0f, 0.0f, 0.0f, 0.0f));
+ spec.x = Max(spec.x, Float4(0.0f));
+ spec.y = Max(spec.y, Float4(0.0f));
+ spec.z = Max(spec.z, Float4(0.0f));
- r.ox[D1] = r.ox[D1] + spec.x;
- r.oy[D1] = r.oy[D1] + spec.y;
- r.oz[D1] = r.oz[D1] + spec.z;
+ r.o[D1].x = r.o[D1].x + spec.x;
+ r.o[D1].y = r.o[D1].y + spec.y;
+ r.o[D1].z = r.o[D1].z + spec.z;
}
}
Float4 globalAmbient = *Pointer<Float4>(r.data + OFFSET(DrawData,ff.globalAmbient)); // FIXME: Unpack
- r.ox[D0] = r.ox[D0] + globalAmbient.x;
- r.oy[D0] = r.oy[D0] + globalAmbient.y;
- r.oz[D0] = r.oz[D0] + globalAmbient.z;
+ r.o[D0].x = r.o[D0].x + globalAmbient.x;
+ r.o[D0].y = r.o[D0].y + globalAmbient.y;
+ r.o[D0].z = r.o[D0].z + globalAmbient.z;
if(state.vertexAmbientMaterialSourceActive == Context::MATERIAL)
{
Float4 materialAmbient = *Pointer<Float4>(r.data + OFFSET(DrawData,ff.materialAmbient)); // FIXME: Unpack
- r.ox[D0] = r.ox[D0] * materialAmbient.x;
- r.oy[D0] = r.oy[D0] * materialAmbient.y;
- r.oz[D0] = r.oz[D0] * materialAmbient.z;
+ r.o[D0].x = r.o[D0].x * materialAmbient.x;
+ r.o[D0].y = r.o[D0].y * materialAmbient.y;
+ r.o[D0].z = r.o[D0].z * materialAmbient.z;
}
else if(state.vertexAmbientMaterialSourceActive == Context::COLOR1)
{
- Color4f materialDiffuse = r.v[Color0];
+ Vector4f materialDiffuse = r.v[Color0];
- r.ox[D0] = r.ox[D0] * materialDiffuse.x;
- r.oy[D0] = r.oy[D0] * materialDiffuse.y;
- r.oz[D0] = r.oz[D0] * materialDiffuse.z;
+ r.o[D0].x = r.o[D0].x * materialDiffuse.x;
+ r.o[D0].y = r.o[D0].y * materialDiffuse.y;
+ r.o[D0].z = r.o[D0].z * materialDiffuse.z;
}
else if(state.vertexAmbientMaterialSourceActive == Context::COLOR2)
{
- Color4f materialSpecular = r.v[Color1];
+ Vector4f materialSpecular = r.v[Color1];
- r.ox[D0] = r.ox[D0] * materialSpecular.x;
- r.oy[D0] = r.oy[D0] * materialSpecular.y;
- r.oz[D0] = r.oz[D0] * materialSpecular.z;
+ r.o[D0].x = r.o[D0].x * materialSpecular.x;
+ r.o[D0].y = r.o[D0].y * materialSpecular.y;
+ r.o[D0].z = r.o[D0].z * materialSpecular.z;
}
else ASSERT(false);
- r.ox[D0] = r.ox[D0] + diffuseSum.x;
- r.oy[D0] = r.oy[D0] + diffuseSum.y;
- r.oz[D0] = r.oz[D0] + diffuseSum.z;
+ r.o[D0].x = r.o[D0].x + diffuseSum.x;
+ r.o[D0].y = r.o[D0].y + diffuseSum.y;
+ r.o[D0].z = r.o[D0].z + diffuseSum.z;
// Emissive
if(state.vertexEmissiveMaterialSourceActive == Context::MATERIAL)
{
Float4 materialEmission = *Pointer<Float4>(r.data + OFFSET(DrawData,ff.materialEmission)); // FIXME: Unpack
- r.ox[D0] = r.ox[D0] + materialEmission.x;
- r.oy[D0] = r.oy[D0] + materialEmission.y;
- r.oz[D0] = r.oz[D0] + materialEmission.z;
+ r.o[D0].x = r.o[D0].x + materialEmission.x;
+ r.o[D0].y = r.o[D0].y + materialEmission.y;
+ r.o[D0].z = r.o[D0].z + materialEmission.z;
}
else if(state.vertexEmissiveMaterialSourceActive == Context::COLOR1)
{
- Color4f materialSpecular = r.v[Color0];
+ Vector4f materialSpecular = r.v[Color0];
- r.ox[D0] = r.ox[D0] + materialSpecular.x;
- r.oy[D0] = r.oy[D0] + materialSpecular.y;
- r.oz[D0] = r.oz[D0] + materialSpecular.z;
+ r.o[D0].x = r.o[D0].x + materialSpecular.x;
+ r.o[D0].y = r.o[D0].y + materialSpecular.y;
+ r.o[D0].z = r.o[D0].z + materialSpecular.z;
}
else if(state.vertexEmissiveMaterialSourceActive == Context::COLOR2)
{
- Color4f materialSpecular = r.v[Color1];
+ Vector4f materialSpecular = r.v[Color1];
- r.ox[D0] = r.ox[D0] + materialSpecular.x;
- r.oy[D0] = r.oy[D0] + materialSpecular.y;
- r.oz[D0] = r.oz[D0] + materialSpecular.z;
+ r.o[D0].x = r.o[D0].x + materialSpecular.x;
+ r.o[D0].y = r.o[D0].y + materialSpecular.y;
+ r.o[D0].z = r.o[D0].z + materialSpecular.z;
}
else ASSERT(false);
// Diffuse alpha component
if(state.vertexDiffuseMaterialSourceActive == Context::MATERIAL)
{
- r.ow[D0] = Float4(*Pointer<Float4>(r.data + OFFSET(DrawData,ff.materialDiffuse[0]))).wwww; // FIXME: Unpack
+ r.o[D0].w = Float4(*Pointer<Float4>(r.data + OFFSET(DrawData,ff.materialDiffuse[0]))).wwww; // FIXME: Unpack
}
else if(state.vertexDiffuseMaterialSourceActive == Context::COLOR1)
{
- Color4f alpha = r.v[Color0];
- r.ow[D0] = alpha.w;
+ Vector4f alpha = r.v[Color0];
+ r.o[D0].w = alpha.w;
}
else if(state.vertexDiffuseMaterialSourceActive == Context::COLOR2)
{
- Color4f alpha = r.v[Color1];
- r.ow[D0] = alpha.w;
+ Vector4f alpha = r.v[Color1];
+ r.o[D0].w = alpha.w;
}
else ASSERT(false);
@@ -469,17 +470,17 @@
// Specular alpha component
if(state.vertexSpecularMaterialSourceActive == Context::MATERIAL)
{
- r.ow[D1] = Float4(*Pointer<Float4>(r.data + OFFSET(DrawData,ff.materialSpecular[3]))).wwww; // FIXME: Unpack
+ r.o[D1].w = Float4(*Pointer<Float4>(r.data + OFFSET(DrawData,ff.materialSpecular[3]))).wwww; // FIXME: Unpack
}
else if(state.vertexSpecularMaterialSourceActive == Context::COLOR1)
{
- Color4f alpha = r.v[Color0];
- r.ow[D1] = alpha.w;
+ Vector4f alpha = r.v[Color0];
+ r.o[D1].w = alpha.w;
}
else if(state.vertexSpecularMaterialSourceActive == Context::COLOR2)
{
- Color4f alpha = r.v[Color1];
- r.ow[D1] = alpha.w;
+ Vector4f alpha = r.v[Color1];
+ r.o[D1].w = alpha.w;
}
else ASSERT(false);
}
@@ -492,11 +493,11 @@
case Context::FOG_NONE:
if(state.specularActive)
{
- r.ox[Fog] = r.ow[D1];
+ r.o[Fog].x = r.o[D1].w;
}
else
{
- r.ox[Fog] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ r.o[Fog].x = Float4(0.0f);
}
break;
case Context::FOG_LINEAR:
@@ -504,21 +505,21 @@
case Context::FOG_EXP2:
if(!state.rangeFogActive)
{
- r.ox[Fog] = r.oz[Pos];
+ r.o[Fog].x = r.o[Pos].z;
}
else
{
- Color4f pos;
+ Vector4f pos;
- pos.x = r.ox[Pos];
- pos.y = r.oy[Pos];
- pos.z = r.oz[Pos];
- pos.w = r.ow[Pos];
+ pos.x = r.o[Pos].x;
+ pos.y = r.o[Pos].y;
+ pos.z = r.o[Pos].z;
+ pos.w = r.o[Pos].w;
- r.ox[Fog] = Sqrt(dot3(pos, pos)); // FIXME: oFog = length(o[Pos]);
+ r.o[Fog].x = Sqrt(dot3(pos, pos)); // FIXME: oFog = length(o[Pos]);
}
- r.ox[Fog] = r.ox[Fog] * *Pointer<Float4>(r.data + OFFSET(DrawData,fog.scale)) + *Pointer<Float4>(r.data + OFFSET(DrawData,fog.offset));
+ r.o[Fog].x = r.o[Fog].x * *Pointer<Float4>(r.data + OFFSET(DrawData,fog.scale)) + *Pointer<Float4>(r.data + OFFSET(DrawData,fog.offset));
break;
default:
ASSERT(false);
@@ -533,7 +534,7 @@
processPointSize(r);
}
- void VertexPipeline::processTextureCoordinate(Registers &r, int stage, Color4f &normal, Color4f &position)
+ void VertexPipeline::processTextureCoordinate(Registers &r, int stage, Vector4f &normal, Vector4f &position)
{
if(state.output[T0 + stage].write)
{
@@ -543,28 +544,28 @@
{
case Context::TEXGEN_PASSTHRU:
{
- Color4f v = r.v[TexCoord0 + i];
+ Vector4f v = r.v[TexCoord0 + i];
- r.ox[T0 + stage] = v.x;
- r.oy[T0 + stage] = v.y;
- r.oz[T0 + stage] = v.z;
- r.ow[T0 + stage] = v.w;
+ r.o[T0 + stage].x = v.x;
+ r.o[T0 + stage].y = v.y;
+ r.o[T0 + stage].z = v.z;
+ r.o[T0 + stage].w = v.w;
- if(state.input[TexCoord0 + i].type == STREAMTYPE_FLOAT)
+ if(state.input[TexCoord0 + i])
{
switch(state.input[TexCoord0 + i].count)
{
case 1:
- r.oy[T0 + stage] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- r.oz[T0 + stage] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- r.ow[T0 + stage] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ r.o[T0 + stage].y = Float4(1.0f);
+ r.o[T0 + stage].z = Float4(0.0f);
+ r.o[T0 + stage].w = Float4(0.0f);
break;
case 2:
- r.oz[T0 + stage] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- r.ow[T0 + stage] = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ r.o[T0 + stage].z = Float4(1.0f);
+ r.o[T0 + stage].w = Float4(0.0f);
break;
case 3:
- r.ow[T0 + stage] = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ r.o[T0 + stage].w = Float4(1.0f);
break;
case 4:
break;
@@ -572,12 +573,11 @@
ASSERT(false);
}
}
- else ASSERT(!state.input[TexCoord0 + i]); // Point sprite; coordinates provided by setup
}
break;
case Context::TEXGEN_NORMAL:
{
- Color4f Nc; // Normal vector in camera space
+ Vector4f Nc; // Normal vector in camera space
if(state.vertexNormalActive)
{
@@ -585,51 +585,51 @@
}
else
{
- Nc.x = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- Nc.y = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- Nc.z = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ Nc.x = Float4(0.0f);
+ Nc.y = Float4(0.0f);
+ Nc.z = Float4(0.0f);
}
- Nc.w = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ Nc.w = Float4(1.0f);
- r.ox[T0 + stage] = Nc.x;
- r.oy[T0 + stage] = Nc.y;
- r.oz[T0 + stage] = Nc.z;
- r.ow[T0 + stage] = Nc.w;
+ r.o[T0 + stage].x = Nc.x;
+ r.o[T0 + stage].y = Nc.y;
+ r.o[T0 + stage].z = Nc.z;
+ r.o[T0 + stage].w = Nc.w;
}
break;
case Context::TEXGEN_POSITION:
{
- Color4f Pn = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true); // Position in camera space
+ Vector4f Pn = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true); // Position in camera space
- Pn.w = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ Pn.w = Float4(1.0f);
- r.ox[T0 + stage] = Pn.x;
- r.oy[T0 + stage] = Pn.y;
- r.oz[T0 + stage] = Pn.z;
- r.ow[T0 + stage] = Pn.w;
+ r.o[T0 + stage].x = Pn.x;
+ r.o[T0 + stage].y = Pn.y;
+ r.o[T0 + stage].z = Pn.z;
+ r.o[T0 + stage].w = Pn.w;
}
break;
case Context::TEXGEN_REFLECTION:
{
- Color4f R; // Reflection vector
+ Vector4f R; // Reflection vector
if(state.vertexNormalActive)
{
- Color4f Nc; // Normal vector in camera space
+ Vector4f Nc; // Normal vector in camera space
Nc = normal;
if(state.localViewerActive)
{
- Color4f Ec; // Eye vector in camera space
- Color4f N2;
+ Vector4f Ec; // Eye vector in camera space
+ Vector4f N2;
Ec = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true);
Ec = normalize(Ec);
// R = E - 2 * N * (E . N)
- Float4 dot = Float4(2.0f, 2.0f, 2.0f, 2.0f) * dot3(Ec, Nc);
+ Float4 dot = Float4(2.0f) * dot3(Ec, Nc);
R.x = Ec.x - Nc.x * dot;
R.y = Ec.y - Nc.y * dot;
@@ -641,46 +641,46 @@
// v = -2 * Nz * Ny
// w = 1 - 2 * Nz * Nz
- R.x = -Float4(2.0f, 2.0f, 2.0f, 2.0f) * Nc.z * Nc.x;
- R.y = -Float4(2.0f, 2.0f, 2.0f, 2.0f) * Nc.z * Nc.y;
- R.z = Float4(1.0f, 1.0f, 1.0f, 1.0f) - Float4(2.0f, 2.0f, 2.0f, 2.0f) * Nc.z * Nc.z;
+ R.x = -Float4(2.0f) * Nc.z * Nc.x;
+ R.y = -Float4(2.0f) * Nc.z * Nc.y;
+ R.z = Float4(1.0f) - Float4(2.0f) * Nc.z * Nc.z;
}
}
else
{
- R.x = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- R.y = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- R.z = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ R.x = Float4(0.0f);
+ R.y = Float4(0.0f);
+ R.z = Float4(0.0f);
}
- R.w = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ R.w = Float4(1.0f);
- r.ox[T0 + stage] = R.x;
- r.oy[T0 + stage] = R.y;
- r.oz[T0 + stage] = R.z;
- r.ow[T0 + stage] = R.w;
+ r.o[T0 + stage].x = R.x;
+ r.o[T0 + stage].y = R.y;
+ r.o[T0 + stage].z = R.z;
+ r.o[T0 + stage].w = R.w;
}
break;
case Context::TEXGEN_SPHEREMAP:
{
- Color4f R; // Reflection vector
+ Vector4f R; // Reflection vector
if(state.vertexNormalActive)
{
- Color4f Nc; // Normal vector in camera space
+ Vector4f Nc; // Normal vector in camera space
Nc = normal;
if(state.localViewerActive)
{
- Color4f Ec; // Eye vector in camera space
- Color4f N2;
+ Vector4f Ec; // Eye vector in camera space
+ Vector4f N2;
Ec = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true);
Ec = normalize(Ec);
// R = E - 2 * N * (E . N)
- Float4 dot = Float4(2.0f, 2.0f, 2.0f, 2.0f) * dot3(Ec, Nc);
+ Float4 dot = Float4(2.0f) * dot3(Ec, Nc);
R.x = Ec.x - Nc.x * dot;
R.y = Ec.y - Nc.y * dot;
@@ -692,48 +692,48 @@
// v = -2 * Nz * Ny
// w = 1 - 2 * Nz * Nz
- R.x = -Float4(2.0f, 2.0f, 2.0f, 2.0f) * Nc.z * Nc.x;
- R.y = -Float4(2.0f, 2.0f, 2.0f, 2.0f) * Nc.z * Nc.y;
- R.z = Float4(1.0f, 1.0f, 1.0f, 1.0f) - Float4(2.0f, 2.0f, 2.0f, 2.0f) * Nc.z * Nc.z;
+ R.x = -Float4(2.0f) * Nc.z * Nc.x;
+ R.y = -Float4(2.0f) * Nc.z * Nc.y;
+ R.z = Float4(1.0f) - Float4(2.0f) * Nc.z * Nc.z;
}
}
else
{
- R.x = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- R.y = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- R.z = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ R.x = Float4(0.0f);
+ R.y = Float4(0.0f);
+ R.z = Float4(0.0f);
}
- R.z -= Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ R.z -= Float4(1.0f);
R = normalize(R);
- R.x = Float4(0.5f, 0.5f, 0.5f, 0.5f) * R.x + Float4(0.5f, 0.5f, 0.5f, 0.5f);
- R.y = Float4(0.5f, 0.5f, 0.5f, 0.5f) * R.y + Float4(0.5f, 0.5f, 0.5f, 0.5f);
+ R.x = Float4(0.5f) * R.x + Float4(0.5f);
+ R.y = Float4(0.5f) * R.y + Float4(0.5f);
- R.z = Float4(1.0f, 1.0f, 1.0f, 1.0f);
- R.w = Float4(0.0f, 0.0f, 0.0f, 0.0f);
+ R.z = Float4(1.0f);
+ R.w = Float4(0.0f);
- r.ox[T0 + stage] = R.x;
- r.oy[T0 + stage] = R.y;
- r.oz[T0 + stage] = R.z;
- r.ow[T0 + stage] = R.w;
+ r.o[T0 + stage].x = R.x;
+ r.o[T0 + stage].y = R.y;
+ r.o[T0 + stage].z = R.z;
+ r.o[T0 + stage].w = R.w;
}
break;
default:
ASSERT(false);
}
- Color4f texTrans0;
- Color4f texTrans1;
- Color4f texTrans2;
- Color4f texTrans3;
+ Vector4f texTrans0;
+ Vector4f texTrans1;
+ Vector4f texTrans2;
+ Vector4f texTrans3;
- Color4f T;
- Color4f t;
+ Vector4f T;
+ Vector4f t;
- T.x = r.ox[T0 + stage];
- T.y = r.oy[T0 + stage];
- T.z = r.oz[T0 + stage];
- T.w = r.ow[T0 + stage];
+ T.x = r.o[T0 + stage].x;
+ T.y = r.o[T0 + stage].y;
+ T.z = r.o[T0 + stage].z;
+ T.w = r.o[T0 + stage].w;
switch(state.textureState[stage].textureTransformCountActive)
{
@@ -766,10 +766,10 @@
texTrans0.w = texTrans0.w.wwww;
t.x = dot4(T, texTrans0);
- r.ox[T0 + stage] = t.x;
- r.oy[T0 + stage] = t.y;
- r.oz[T0 + stage] = t.z;
- r.ow[T0 + stage] = t.w;
+ r.o[T0 + stage].x = t.x;
+ r.o[T0 + stage].y = t.y;
+ r.o[T0 + stage].z = t.z;
+ r.o[T0 + stage].w = t.w;
case 0:
break;
default:
@@ -787,16 +787,16 @@
if(state.input[PSize])
{
- r.oy[Pts] = r.v[PSize].x;
+ r.o[Pts].y = r.v[PSize].x;
}
else
{
- r.oy[Pts] = *Pointer<Float4>(r.data + OFFSET(DrawData,point.pointSize));
+ r.o[Pts].y = *Pointer<Float4>(r.data + OFFSET(DrawData,point.pointSize));
}
if(state.pointScaleActive && !state.preTransformed)
{
- Color4f p = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true);
+ Vector4f p = transformBlend(r, r.v[Position], Pointer<Byte>(r.data + OFFSET(DrawData,ff.cameraTransformT)), true);
Float4 d = Sqrt(dot3(p, p)); // FIXME: length(p);
@@ -806,18 +806,13 @@
A = RcpSqrt_pp(A + d * (B + d * C));
- r.oy[Pts] = r.oy[Pts] * Float4(*Pointer<Float>(r.data + OFFSET(DrawData,viewportHeight))) * A; // FIXME: Unpack
+ r.o[Pts].y = r.o[Pts].y * Float4(*Pointer<Float>(r.data + OFFSET(DrawData,viewportHeight))) * A; // FIXME: Unpack
}
}
- Color4f VertexPipeline::transform(Color4f &src, Pointer<Byte> &matrix, bool homogeneous)
+ Vector4f VertexPipeline::transform(Register &src, Pointer<Byte> &matrix, bool homogeneous)
{
- Color4f dst;
-
- Color4f row0;
- Color4f row1;
- Color4f row2;
- Color4f row3;
+ Vector4f dst;
if(homogeneous)
{
@@ -862,14 +857,9 @@
return dst;
}
- Color4f VertexPipeline::transform(Color4f &src, Pointer<Byte> &matrix, UInt index[4], bool homogeneous)
+ Vector4f VertexPipeline::transform(Register &src, Pointer<Byte> &matrix, UInt index[4], bool homogeneous)
{
- Color4f dst;
-
- Color4f row0;
- Color4f row1;
- Color4f row2;
- Color4f row3;
+ Vector4f dst;
if(homogeneous)
{
@@ -914,9 +904,9 @@
return dst;
}
- Color4f VertexPipeline::normalize(Color4f &src)
+ Vector4f VertexPipeline::normalize(Vector4f &src)
{
- Color4f dst;
+ Vector4f dst;
Float4 rcpLength = RcpSqrt_pp(dot3(src, src));
@@ -933,11 +923,11 @@
dst = dst * dst;
dst = dst * dst;
- dst = Float4(As<Int4>(dst) - As<Int4>(Float4(1.0f, 1.0f, 1.0f, 1.0f)));
+ dst = Float4(As<Int4>(dst) - As<Int4>(Float4(1.0f)));
dst *= src1;
- dst = As<Float4>(Int4(dst) + As<Int4>(Float4(1.0f, 1.0f, 1.0f, 1.0f)));
+ dst = As<Float4>(Int4(dst) + As<Int4>(Float4(1.0f)));
dst = RcpSqrt_pp(dst);
dst = RcpSqrt_pp(dst);
diff --git a/src/Shader/VertexPipeline.hpp b/src/Shader/VertexPipeline.hpp
index efb8592..bfc3093 100644
--- a/src/Shader/VertexPipeline.hpp
+++ b/src/Shader/VertexPipeline.hpp
@@ -1,42 +1,42 @@
-// SwiftShader Software Renderer
-//
-// Copyright(c) 2005-2011 TransGaming Inc.
-//
-// All rights reserved. No part of this software may be copied, distributed, transmitted,
-// transcribed, stored in a retrieval system, translated into any human or computer
-// language by any means, or disclosed to third parties without the explicit written
-// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
-// or implied, including but not limited to any patent rights, are granted to you.
-//
-
-#ifndef sw_VertexPipeline_hpp
-#define sw_VertexPipeline_hpp
-
-#include "VertexRoutine.hpp"
-
-#include "Context.hpp"
-#include "VertexProcessor.hpp"
-
-namespace sw
-{
- class VertexPipeline : public VertexRoutine
- {
- public:
- VertexPipeline(const VertexProcessor::State &state);
-
- virtual ~VertexPipeline();
-
- private:
- void pipeline(Registers &r);
- void processTextureCoordinate(Registers &r, int stage, Color4f &normal, Color4f &position);
- void processPointSize(Registers &r);
-
- Color4f transformBlend(Registers &r, Color4f &src, Pointer<Byte> &matrix, bool homogenous);
- Color4f transform(Color4f &src, Pointer<Byte> &matrix, bool homogenous);
- Color4f transform(Color4f &src, Pointer<Byte> &matrix, UInt index[4], bool homogenous);
- Color4f normalize(Color4f &src);
- Float4 power(Float4 &src0, Float4 &src1);
- };
-};
-
-#endif // sw_VertexPipeline_hpp
+// SwiftShader Software Renderer
+//
+// Copyright(c) 2005-2012 TransGaming Inc.
+//
+// All rights reserved. No part of this software may be copied, distributed, transmitted,
+// transcribed, stored in a retrieval system, translated into any human or computer
+// language by any means, or disclosed to third parties without the explicit written
+// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
+// or implied, including but not limited to any patent rights, are granted to you.
+//
+
+#ifndef sw_VertexPipeline_hpp
+#define sw_VertexPipeline_hpp
+
+#include "VertexRoutine.hpp"
+
+#include "Context.hpp"
+#include "VertexProcessor.hpp"
+
+namespace sw
+{
+ class VertexPipeline : public VertexRoutine
+ {
+ public:
+ VertexPipeline(const VertexProcessor::State &state);
+
+ virtual ~VertexPipeline();
+
+ private:
+ void pipeline(Registers &r);
+ void processTextureCoordinate(Registers &r, int stage, Vector4f &normal, Vector4f &position);
+ void processPointSize(Registers &r);
+
+ Vector4f transformBlend(Registers &r, Register &src, Pointer<Byte> &matrix, bool homogenous);
+ Vector4f transform(Register &src, Pointer<Byte> &matrix, bool homogenous);
+ Vector4f transform(Register &src, Pointer<Byte> &matrix, UInt index[4], bool homogenous);
+ Vector4f normalize(Vector4f &src);
+ Float4 power(Float4 &src0, Float4 &src1);
+ };
+};
+
+#endif // sw_VertexPipeline_hpp
diff --git a/src/Shader/VertexProgram.cpp b/src/Shader/VertexProgram.cpp
index 41153bf..a292a8e 100644
--- a/src/Shader/VertexProgram.cpp
+++ b/src/Shader/VertexProgram.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -22,12 +22,13 @@
namespace sw
{
- VertexProgram::VertexProgram(const VertexProcessor::State &state, const VertexShader *vertexShader) : VertexRoutine(state), vertexShader(vertexShader)
+ VertexProgram::VertexProgram(const VertexProcessor::State &state, const VertexShader *shader) : VertexRoutine(state, shader)
{
- returns = false;
ifDepth = 0;
loopRepDepth = 0;
breakDepth = 0;
+ currentLabel = -1;
+ whileTest = false;
for(int i = 0; i < 2048; i++)
{
@@ -52,7 +53,7 @@
if(!state.preTransformed)
{
- shader(r);
+ program(r);
}
else
{
@@ -60,296 +61,320 @@
}
}
- Color4f VertexProgram::readConstant(Registers &r, const Src &src, int offset)
+ void VertexProgram::program(Registers &r)
{
- Color4f c;
+ // shader->print("VertexShader-%0.8X.txt", state.shaderID);
- int i = src.index + offset;
- bool relative = src.relative;
-
- if(!relative)
- {
- c.r = c.g = c.b = c.a = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c[i]));
-
- c.r = c.r.xxxx;
- c.g = c.g.yyyy;
- c.b = c.b.zzzz;
- c.a = c.a.wwww;
-
- if(localShaderConstants) // Constant may be known at compile time
- {
- for(int j = 0; j < vertexShader->getLength(); j++)
- {
- const ShaderInstruction &instruction = *vertexShader->getInstruction(j);
-
- if(instruction.getOpcode() == ShaderOperation::OPCODE_DEF)
- {
- if(instruction.getDestinationParameter().index == i)
- {
- c.r = Float4(instruction.getSourceParameter(0).value);
- c.g = Float4(instruction.getSourceParameter(1).value);
- c.b = Float4(instruction.getSourceParameter(2).value);
- c.a = Float4(instruction.getSourceParameter(3).value);
-
- break;
- }
- }
- }
- }
- }
- else if(src.relativeType == Src::PARAMETER_LOOP)
- {
- Int loopCounter = r.aL[r.loopDepth];
-
- c.r = c.g = c.b = c.a = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c[i]) + loopCounter * 16);
-
- c.r = c.r.xxxx;
- c.g = c.g.yyyy;
- c.b = c.b.zzzz;
- c.a = c.a.wwww;
- }
- else
- {
- Int index0;
- Int index1;
- Int index2;
- Int index3;
-
- Float4 a0_;
-
- switch(src.relativeSwizzle & 0x03)
- {
- case 0: a0_ = r.a0.x; break;
- case 1: a0_ = r.a0.y; break;
- case 2: a0_ = r.a0.z; break;
- case 3: a0_ = r.a0.w; break;
- }
-
- index0 = i + RoundInt(Float(a0_.x));
- index1 = i + RoundInt(Float(a0_.y));
- index2 = i + RoundInt(Float(a0_.z));
- index3 = i + RoundInt(Float(a0_.w));
-
- // Clamp to constant register range, c[256] = {0, 0, 0, 0}
- index0 = IfThenElse(UInt(index0) > UInt(256), Int(256), index0);
- index1 = IfThenElse(UInt(index1) > UInt(256), Int(256), index1);
- index2 = IfThenElse(UInt(index2) > UInt(256), Int(256), index2);
- index3 = IfThenElse(UInt(index3) > UInt(256), Int(256), index3);
-
- c.x = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index0 * 16, 16);
- c.y = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index1 * 16, 16);
- c.z = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index2 * 16, 16);
- c.w = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index3 * 16, 16);
-
- transpose4x4(c.x, c.y, c.z, c.w);
- }
-
- return c;
- }
-
- void VertexProgram::shader(Registers &r)
- {
- // vertexShader->print("VertexShader-%0.16llX.txt", state.shaderHash);
-
- unsigned short version = vertexShader->getVersion();
+ unsigned short version = shader->getVersion();
r.enableIndex = 0;
r.stackIndex = 0;
-
- for(int i = 0; i < vertexShader->getLength(); i++)
+
+ // Create all call site return blocks up front
+ for(int i = 0; i < shader->getLength(); i++)
{
- const ShaderInstruction *instruction = vertexShader->getInstruction(i);
- Op::Opcode opcode = instruction->getOpcode();
+ const Shader::Instruction *instruction = shader->getInstruction(i);
+ Shader::Opcode opcode = instruction->opcode;
- // #ifndef NDEBUG // FIXME: Centralize debug output control
- // vertexShader->printInstruction(i, "debug.txt");
- // #endif
+ if(opcode == Shader::OPCODE_CALL || opcode == Shader::OPCODE_CALLNZ)
+ {
+ const Dst &dst = instruction->dst;
- if(opcode == Op::OPCODE_DCL || opcode == Op::OPCODE_DEF || opcode == Op::OPCODE_DEFI || opcode == Op::OPCODE_DEFB)
+ ASSERT(callRetBlock[dst.label].size() == dst.callSite);
+ callRetBlock[dst.label].push_back(Nucleus::createBasicBlock());
+ }
+ }
+
+ for(int i = 0; i < shader->getLength(); i++)
+ {
+ const Shader::Instruction *instruction = shader->getInstruction(i);
+ Shader::Opcode opcode = instruction->opcode;
+
+ if(opcode == Shader::OPCODE_DCL || opcode == Shader::OPCODE_DEF || opcode == Shader::OPCODE_DEFI || opcode == Shader::OPCODE_DEFB)
{
continue;
}
- Dst dest = instruction->getDestinationParameter();
- Src src0 = instruction->getSourceParameter(0);
- Src src1 = instruction->getSourceParameter(1);
- Src src2 = instruction->getSourceParameter(2);
- Src src3 = instruction->getSourceParameter(3);
+ Dst dst = instruction->dst;
+ Src src0 = instruction->src[0];
+ Src src1 = instruction->src[1];
+ Src src2 = instruction->src[2];
- bool predicate = instruction->isPredicate();
- int size = vertexShader->size(opcode);
- Usage usage = instruction->getUsage();
- unsigned char usageIndex = instruction->getUsageIndex();
- Control control = instruction->getControl();
- bool integer = dest.type == Dst::PARAMETER_ADDR;
- bool pp = dest.partialPrecision;
+ bool predicate = instruction->predicate;
+ int size = shader->size(opcode);
+ Usage usage = instruction->usage;
+ unsigned char usageIndex = instruction->usageIndex;
+ Control control = instruction->control;
+ bool integer = dst.type == Shader::PARAMETER_ADDR;
+ bool pp = dst.partialPrecision;
- Color4f d;
- Color4f s0;
- Color4f s1;
- Color4f s2;
- Color4f s3;
+ Vector4f d;
+ Vector4f s0;
+ Vector4f s1;
+ Vector4f s2;
- if(src0.type != Src::PARAMETER_VOID) s0 = reg(r, src0);
- if(src1.type != Src::PARAMETER_VOID) s1 = reg(r, src1);
- if(src2.type != Src::PARAMETER_VOID) s2 = reg(r, src2);
- if(src3.type != Src::PARAMETER_VOID) s3 = reg(r, src3);
+ if(src0.type != Shader::PARAMETER_VOID) s0 = reg(r, src0);
+ if(src1.type != Shader::PARAMETER_VOID) s1 = reg(r, src1);
+ if(src2.type != Shader::PARAMETER_VOID) s2 = reg(r, src2);
switch(opcode)
{
- case Op::OPCODE_VS_1_0: break;
- case Op::OPCODE_VS_1_1: break;
- case Op::OPCODE_VS_2_0: break;
- case Op::OPCODE_VS_2_x: break;
- case Op::OPCODE_VS_2_sw: break;
- case Op::OPCODE_VS_3_0: break;
- case Op::OPCODE_VS_3_sw: break;
- case Op::OPCODE_DCL: break;
- case Op::OPCODE_DEF: break;
- case Op::OPCODE_DEFI: break;
- case Op::OPCODE_DEFB: break;
- case Op::OPCODE_NOP: break;
- case Op::OPCODE_ABS: abs(d, s0); break;
- case Op::OPCODE_ADD: add(d, s0, s1); break;
- case Op::OPCODE_CRS: crs(d, s0, s1); break;
- case Op::OPCODE_DP3: dp3(d, s0, s1); break;
- case Op::OPCODE_DP4: dp4(d, s0, s1); break;
- case Op::OPCODE_DST: dst(d, s0, s1); break;
- case Op::OPCODE_EXP: exp(d, s0, pp); break;
- case Op::OPCODE_EXPP: expp(d, s0, version); break;
- case Op::OPCODE_FRC: frc(d, s0); break;
- case Op::OPCODE_LIT: lit(d, s0); break;
- case Op::OPCODE_LOG: log(d, s0, pp); break;
- case Op::OPCODE_LOGP: logp(d, s0, version); break;
- case Op::OPCODE_LRP: lrp(d, s0, s1, s2); break;
- case Op::OPCODE_M3X2: M3X2(r, d, s0, src1); break;
- case Op::OPCODE_M3X3: M3X3(r, d, s0, src1); break;
- case Op::OPCODE_M3X4: M3X4(r, d, s0, src1); break;
- case Op::OPCODE_M4X3: M4X3(r, d, s0, src1); break;
- case Op::OPCODE_M4X4: M4X4(r, d, s0, src1); break;
- case Op::OPCODE_MAD: mad(d, s0, s1, s2); break;
- case Op::OPCODE_MAX: max(d, s0, s1); break;
- case Op::OPCODE_MIN: min(d, s0, s1); break;
- case Op::OPCODE_MOV: mov(d, s0, integer); break;
- case Op::OPCODE_MOVA: mov(d, s0); break;
- case Op::OPCODE_MUL: mul(d, s0, s1); break;
- case Op::OPCODE_NRM: nrm(d, s0, pp); break;
- case Op::OPCODE_POW: pow(d, s0, s1, pp); break;
- case Op::OPCODE_RCP: rcp(d, s0, pp); break;
- case Op::OPCODE_RSQ: rsq(d, s0, pp); break;
- case Op::OPCODE_SGE: sge(d, s0, s1); break;
- case Op::OPCODE_SGN: sgn(d, s0); break;
- case Op::OPCODE_SINCOS: sincos(d, s0, pp); break;
- case Op::OPCODE_SLT: slt(d, s0, s1); break;
- case Op::OPCODE_SUB: sub(d, s0, s1); break;
- case Op::OPCODE_BREAK: BREAK(r); break;
- case Op::OPCODE_BREAKC: BREAKC(r, s0, s1, control); break;
- case Op::OPCODE_BREAKP: BREAKP(r, src0); break;
- case Op::OPCODE_CALL: CALL(r, dest.index); break;
- case Op::OPCODE_CALLNZ: CALLNZ(r, dest.index, src0); break;
- case Op::OPCODE_ELSE: ELSE(r); break;
- case Op::OPCODE_ENDIF: ENDIF(r); break;
- case Op::OPCODE_ENDLOOP: ENDLOOP(r); break;
- case Op::OPCODE_ENDREP: ENDREP(r); break;
- case Op::OPCODE_IF: IF(r, src0); break;
- case Op::OPCODE_IFC: IFC(r, s0, s1, control); break;
- case Op::OPCODE_LABEL: LABEL(dest.index); break;
- case Op::OPCODE_LOOP: LOOP(r, src1); break;
- case Op::OPCODE_REP: REP(r, src0); break;
- case Op::OPCODE_RET: RET(r); break;
- case Op::OPCODE_SETP: setp(d, s0, s1, control); break;
- case Op::OPCODE_TEXLDL: TEXLDL(r, d, s0, src1); break;
- case Op::OPCODE_END: break;
+ case Shader::OPCODE_VS_1_0: break;
+ case Shader::OPCODE_VS_1_1: break;
+ case Shader::OPCODE_VS_2_0: break;
+ case Shader::OPCODE_VS_2_x: break;
+ case Shader::OPCODE_VS_2_sw: break;
+ case Shader::OPCODE_VS_3_0: break;
+ case Shader::OPCODE_VS_3_sw: break;
+ case Shader::OPCODE_DCL: break;
+ case Shader::OPCODE_DEF: break;
+ case Shader::OPCODE_DEFI: break;
+ case Shader::OPCODE_DEFB: break;
+ case Shader::OPCODE_NOP: break;
+ case Shader::OPCODE_ABS: abs(d, s0); break;
+ case Shader::OPCODE_ADD: add(d, s0, s1); break;
+ case Shader::OPCODE_CRS: crs(d, s0, s1); break;
+ case Shader::OPCODE_FORWARD1: forward1(d, s0, s1, s2); break;
+ case Shader::OPCODE_FORWARD2: forward2(d, s0, s1, s2); break;
+ case Shader::OPCODE_FORWARD3: forward3(d, s0, s1, s2); break;
+ case Shader::OPCODE_FORWARD4: forward4(d, s0, s1, s2); break;
+ case Shader::OPCODE_REFLECT1: reflect1(d, s0, s1); break;
+ case Shader::OPCODE_REFLECT2: reflect2(d, s0, s1); break;
+ case Shader::OPCODE_REFLECT3: reflect3(d, s0, s1); break;
+ case Shader::OPCODE_REFLECT4: reflect4(d, s0, s1); break;
+ case Shader::OPCODE_REFRACT1: refract1(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_REFRACT2: refract2(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_REFRACT3: refract3(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_REFRACT4: refract4(d, s0, s1, s2.x); break;
+ case Shader::OPCODE_DP1: dp1(d, s0, s1); break;
+ case Shader::OPCODE_DP2: dp2(d, s0, s1); break;
+ case Shader::OPCODE_DP3: dp3(d, s0, s1); break;
+ case Shader::OPCODE_DP4: dp4(d, s0, s1); break;
+ case Shader::OPCODE_ATT: att(d, s0, s1); break;
+ case Shader::OPCODE_EXP2X: exp2x(d, s0, pp); break;
+ case Shader::OPCODE_EXP2: exp2(d, s0, pp); break;
+ case Shader::OPCODE_EXPP: expp(d, s0, version); break;
+ case Shader::OPCODE_EXP: exp(d, s0, pp); break;
+ case Shader::OPCODE_FRC: frc(d, s0); break;
+ case Shader::OPCODE_TRUNC: trunc(d, s0); break;
+ case Shader::OPCODE_FLOOR: floor(d, s0); break;
+ case Shader::OPCODE_CEIL: ceil(d, s0); break;
+ case Shader::OPCODE_LIT: lit(d, s0); break;
+ case Shader::OPCODE_LOG2X: log2x(d, s0, pp); break;
+ case Shader::OPCODE_LOG2: log2(d, s0, pp); break;
+ case Shader::OPCODE_LOGP: logp(d, s0, version); break;
+ case Shader::OPCODE_LOG: log(d, s0, pp); break;
+ case Shader::OPCODE_LRP: lrp(d, s0, s1, s2); break;
+ case Shader::OPCODE_STEP: step(d, s0, s1); break;
+ case Shader::OPCODE_SMOOTH: smooth(d, s0, s1, s2); break;
+ case Shader::OPCODE_M3X2: M3X2(r, d, s0, src1); break;
+ case Shader::OPCODE_M3X3: M3X3(r, d, s0, src1); break;
+ case Shader::OPCODE_M3X4: M3X4(r, d, s0, src1); break;
+ case Shader::OPCODE_M4X3: M4X3(r, d, s0, src1); break;
+ case Shader::OPCODE_M4X4: M4X4(r, d, s0, src1); break;
+ case Shader::OPCODE_MAD: mad(d, s0, s1, s2); break;
+ case Shader::OPCODE_MAX: max(d, s0, s1); break;
+ case Shader::OPCODE_MIN: min(d, s0, s1); break;
+ case Shader::OPCODE_MOV: mov(d, s0, integer); break;
+ case Shader::OPCODE_MOVA: mov(d, s0); break;
+ case Shader::OPCODE_F2B: f2b(d, s0); break;
+ case Shader::OPCODE_B2F: b2f(d, s0); break;
+ case Shader::OPCODE_MUL: mul(d, s0, s1); break;
+ case Shader::OPCODE_NRM2: nrm2(d, s0, pp); break;
+ case Shader::OPCODE_NRM3: nrm3(d, s0, pp); break;
+ case Shader::OPCODE_NRM4: nrm4(d, s0, pp); break;
+ case Shader::OPCODE_POWX: powx(d, s0, s1, pp); break;
+ case Shader::OPCODE_POW: pow(d, s0, s1, pp); break;
+ case Shader::OPCODE_RCPX: rcpx(d, s0, pp); break;
+ case Shader::OPCODE_DIV: div(d, s0, s1); break;
+ case Shader::OPCODE_MOD: mod(d, s0, s1); break;
+ case Shader::OPCODE_RSQX: rsqx(d, s0, pp); break;
+ case Shader::OPCODE_SQRT: sqrt(d, s0, pp); break;
+ case Shader::OPCODE_RSQ: rsq(d, s0, pp); break;
+ case Shader::OPCODE_LEN2: len2(d.x, s0, pp); break;
+ case Shader::OPCODE_LEN3: len3(d.x, s0, pp); break;
+ case Shader::OPCODE_LEN4: len4(d.x, s0, pp); break;
+ case Shader::OPCODE_DIST1: dist1(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_DIST2: dist2(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_DIST3: dist3(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_DIST4: dist4(d.x, s0, s1, pp); break;
+ case Shader::OPCODE_SGE: step(d, s1, s0); break;
+ case Shader::OPCODE_SGN: sgn(d, s0); break;
+ case Shader::OPCODE_SINCOS: sincos(d, s0, pp); break;
+ case Shader::OPCODE_COS: cos(d, s0, pp); break;
+ case Shader::OPCODE_SIN: sin(d, s0, pp); break;
+ case Shader::OPCODE_TAN: tan(d, s0); break;
+ case Shader::OPCODE_ACOS: acos(d, s0); break;
+ case Shader::OPCODE_ASIN: asin(d, s0); break;
+ case Shader::OPCODE_ATAN: atan(d, s0); break;
+ case Shader::OPCODE_ATAN2: atan2(d, s0, s1); break;
+ case Shader::OPCODE_SLT: slt(d, s0, s1); break;
+ case Shader::OPCODE_SUB: sub(d, s0, s1); 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_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_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_CMP: cmp(d, s0, s1, control); break;
+ case Shader::OPCODE_ICMP: icmp(d, s0, s1, control); break;
+ case Shader::OPCODE_SELECT: select(d, s0, s1, s2); break;
+ case Shader::OPCODE_EXTRACT: extract(d.x, s0, s1.x); break;
+ case Shader::OPCODE_INSERT: insert(d, s0, s1.x, s2.x); break;
+ case Shader::OPCODE_ALL: all(d.x, s0); break;
+ case Shader::OPCODE_ANY: any(d.x, s0); break;
+ case Shader::OPCODE_NOT: not(d, s0); break;
+ case Shader::OPCODE_OR: or(d.x, s0.x, s1.x); break;
+ case Shader::OPCODE_XOR: xor(d.x, s0.x, s1.x); break;
+ case Shader::OPCODE_AND: and(d.x, s0.x, s1.x); break;
+ case Shader::OPCODE_TEXLDL: TEXLDL(r, d, s0, src1); break;
+ case Shader::OPCODE_TEX: TEX(r, d, s0, src1); break;
+ case Shader::OPCODE_END: break;
default:
ASSERT(false);
}
- if(dest.type != Dst::PARAMETER_VOID && dest.type != Dst::PARAMETER_LABEL)
+ if(dst.type != Shader::PARAMETER_VOID && dst.type != Shader::PARAMETER_LABEL && opcode != Shader::OPCODE_NOP)
{
- if(dest.saturate)
+ if(dst.integer)
{
- if(dest.x) d.r = Max(d.r, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- if(dest.y) d.g = Max(d.g, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- if(dest.z) d.b = Max(d.b, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- if(dest.w) d.a = Max(d.a, Float4(0.0f, 0.0f, 0.0f, 0.0f));
-
- if(dest.x) d.r = Min(d.r, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- if(dest.y) d.g = Min(d.g, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- if(dest.z) d.b = Min(d.b, Float4(1.0f, 1.0f, 1.0f, 1.0f));
- if(dest.w) d.a = Min(d.a, Float4(1.0f, 1.0f, 1.0f, 1.0f));
+ switch(opcode)
+ {
+ case Shader::OPCODE_DIV:
+ if(dst.x) d.x = Trunc(d.x);
+ if(dst.y) d.y = Trunc(d.y);
+ if(dst.z) d.z = Trunc(d.z);
+ if(dst.w) d.w = Trunc(d.w);
+ break;
+ default:
+ break; // No truncation to integer required when arguments are integer
+ }
}
- if(vertexShader->containsDynamicBranching())
+ if(dst.saturate)
{
- Color4f pDst; // FIXME: Rename
+ if(dst.x) d.x = Max(d.x, Float4(0.0f));
+ if(dst.y) d.y = Max(d.y, Float4(0.0f));
+ if(dst.z) d.z = Max(d.z, Float4(0.0f));
+ if(dst.w) d.w = Max(d.w, Float4(0.0f));
- switch(dest.type)
+ if(dst.x) d.x = Min(d.x, Float4(1.0f));
+ if(dst.y) d.y = Min(d.y, Float4(1.0f));
+ if(dst.z) d.z = Min(d.z, Float4(1.0f));
+ if(dst.w) d.w = Min(d.w, Float4(1.0f));
+ }
+
+ if(shader->containsDynamicBranching())
+ {
+ Vector4f pDst; // FIXME: Rename
+
+ switch(dst.type)
{
- case Dst::PARAMETER_VOID: break;
- case Dst::PARAMETER_TEMP: pDst = r.r[dest.index]; break;
- case Dst::PARAMETER_ADDR: pDst = r.a0; break;
- case Dst::PARAMETER_RASTOUT:
- switch(dest.index)
+ case Shader::PARAMETER_VOID: break;
+ 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;
+ }
+ else
+ {
+ Int a = relativeAddress(r, 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;
+ }
+ break;
+ case Shader::PARAMETER_ADDR: pDst = r.a0; break;
+ case Shader::PARAMETER_RASTOUT:
+ switch(dst.index)
{
case 0:
- if(dest.x) pDst.x = r.ox[Pos];
- if(dest.y) pDst.y = r.oy[Pos];
- if(dest.z) pDst.z = r.oz[Pos];
- if(dest.w) pDst.w = r.ow[Pos];
+ if(dst.x) pDst.x = r.o[Pos].x;
+ if(dst.y) pDst.y = r.o[Pos].y;
+ if(dst.z) pDst.z = r.o[Pos].z;
+ if(dst.w) pDst.w = r.o[Pos].w;
break;
case 1:
- pDst.x = r.ox[Fog];
+ pDst.x = r.o[Fog].x;
break;
case 2:
- pDst.x = r.oy[Pts];
+ pDst.x = r.o[Pts].y;
break;
default:
ASSERT(false);
}
break;
- case Dst::PARAMETER_ATTROUT:
- if(dest.x) pDst.x = r.ox[D0 + dest.index];
- if(dest.y) pDst.y = r.oy[D0 + dest.index];
- if(dest.z) pDst.z = r.oz[D0 + dest.index];
- if(dest.w) pDst.w = r.ow[D0 + dest.index];
+ case Shader::PARAMETER_ATTROUT:
+ if(dst.x) pDst.x = r.o[D0 + dst.index].x;
+ if(dst.y) pDst.y = r.o[D0 + dst.index].y;
+ if(dst.z) pDst.z = r.o[D0 + dst.index].z;
+ if(dst.w) pDst.w = r.o[D0 + dst.index].w;
break;
- case Dst::PARAMETER_TEXCRDOUT:
- // case Dst::PARAMETER_OUTPUT:
+ case Shader::PARAMETER_TEXCRDOUT:
+ // case Shader::PARAMETER_OUTPUT:
if(version < 0x0300)
{
- if(dest.x) pDst.x = r.ox[T0 + dest.index];
- if(dest.y) pDst.y = r.oy[T0 + dest.index];
- if(dest.z) pDst.z = r.oz[T0 + dest.index];
- if(dest.w) pDst.w = r.ow[T0 + dest.index];
+ if(dst.x) pDst.x = r.o[T0 + dst.index].x;
+ if(dst.y) pDst.y = r.o[T0 + dst.index].y;
+ if(dst.z) pDst.z = r.o[T0 + dst.index].z;
+ if(dst.w) pDst.w = r.o[T0 + dst.index].w;
}
else
{
- if(!dest.relative)
+ if(dst.rel.type == Shader::PARAMETER_VOID) // Not relative
{
- if(dest.x) pDst.x = r.ox[dest.index];
- if(dest.y) pDst.y = r.oy[dest.index];
- if(dest.z) pDst.z = r.oz[dest.index];
- if(dest.w) pDst.w = r.ow[dest.index];
+ if(dst.x) pDst.x = r.o[dst.index].x;
+ if(dst.y) pDst.y = r.o[dst.index].y;
+ if(dst.z) pDst.z = r.o[dst.index].z;
+ if(dst.w) pDst.w = r.o[dst.index].w;
}
- else
+ else if(dst.rel.type == Shader::PARAMETER_LOOP)
{
Int aL = r.aL[r.loopDepth];
- if(dest.x) pDst.x = r.ox[dest.index + aL];
- if(dest.y) pDst.y = r.oy[dest.index + aL];
- if(dest.z) pDst.z = r.oz[dest.index + aL];
- if(dest.w) pDst.w = r.ow[dest.index + aL];
+ if(dst.x) pDst.x = r.o[dst.index + aL].x;
+ if(dst.y) pDst.y = r.o[dst.index + aL].y;
+ if(dst.z) pDst.z = r.o[dst.index + aL].z;
+ if(dst.w) pDst.w = r.o[dst.index + aL].w;
+ }
+ else
+ {
+ Int a = relativeAddress(r, dst);
+
+ if(dst.x) pDst.x = r.o[dst.index + a].x;
+ if(dst.y) pDst.y = r.o[dst.index + a].y;
+ if(dst.z) pDst.z = r.o[dst.index + a].z;
+ if(dst.w) pDst.w = r.o[dst.index + a].w;
}
}
break;
- case Dst::PARAMETER_LABEL: break;
- case Dst::PARAMETER_PREDICATE: pDst = r.p0; break;
- case Dst::PARAMETER_INPUT: break;
+ case Shader::PARAMETER_LABEL: break;
+ case Shader::PARAMETER_PREDICATE: pDst = r.p0; break;
+ case Shader::PARAMETER_INPUT: break;
default:
ASSERT(false);
}
- Int4 enable = r.enableStack[r.enableIndex] & r.enableBreak;
+ Int4 enable = enableMask(r, instruction);
Int4 xEnable = enable;
Int4 yEnable = enable;
@@ -358,119 +383,140 @@
if(predicate)
{
- unsigned char pSwizzle = instruction->getPredicateSwizzle();
+ 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];
- if(!instruction->isPredicateNot())
+ if(!instruction->predicateNot)
{
- if(dest.x) xEnable = xEnable & As<Int4>(xPredicate);
- if(dest.y) yEnable = yEnable & As<Int4>(yPredicate);
- if(dest.z) zEnable = zEnable & As<Int4>(zPredicate);
- if(dest.w) wEnable = wEnable & As<Int4>(wPredicate);
+ if(dst.x) xEnable = xEnable & As<Int4>(xPredicate);
+ if(dst.y) yEnable = yEnable & As<Int4>(yPredicate);
+ if(dst.z) zEnable = zEnable & As<Int4>(zPredicate);
+ if(dst.w) wEnable = wEnable & As<Int4>(wPredicate);
}
else
{
- if(dest.x) xEnable = xEnable & ~As<Int4>(xPredicate);
- if(dest.y) yEnable = yEnable & ~As<Int4>(yPredicate);
- if(dest.z) zEnable = zEnable & ~As<Int4>(zPredicate);
- if(dest.w) wEnable = wEnable & ~As<Int4>(wPredicate);
+ if(dst.x) xEnable = xEnable & ~As<Int4>(xPredicate);
+ if(dst.y) yEnable = yEnable & ~As<Int4>(yPredicate);
+ if(dst.z) zEnable = zEnable & ~As<Int4>(zPredicate);
+ if(dst.w) wEnable = wEnable & ~As<Int4>(wPredicate);
}
}
- if(dest.x) d.x = As<Float4>(As<Int4>(d.x) & xEnable);
- if(dest.y) d.y = As<Float4>(As<Int4>(d.y) & yEnable);
- if(dest.z) d.z = As<Float4>(As<Int4>(d.z) & zEnable);
- if(dest.w) d.w = As<Float4>(As<Int4>(d.w) & wEnable);
+ if(dst.x) d.x = As<Float4>(As<Int4>(d.x) & xEnable);
+ if(dst.y) d.y = As<Float4>(As<Int4>(d.y) & yEnable);
+ if(dst.z) d.z = As<Float4>(As<Int4>(d.z) & zEnable);
+ if(dst.w) d.w = As<Float4>(As<Int4>(d.w) & wEnable);
- if(dest.x) d.x = As<Float4>(As<Int4>(d.x) | (As<Int4>(pDst.x) & ~xEnable));
- if(dest.y) d.y = As<Float4>(As<Int4>(d.y) | (As<Int4>(pDst.y) & ~yEnable));
- if(dest.z) d.z = As<Float4>(As<Int4>(d.z) | (As<Int4>(pDst.z) & ~zEnable));
- if(dest.w) d.w = As<Float4>(As<Int4>(d.w) | (As<Int4>(pDst.w) & ~wEnable));
+ if(dst.x) d.x = As<Float4>(As<Int4>(d.x) | (As<Int4>(pDst.x) & ~xEnable));
+ if(dst.y) d.y = As<Float4>(As<Int4>(d.y) | (As<Int4>(pDst.y) & ~yEnable));
+ if(dst.z) d.z = As<Float4>(As<Int4>(d.z) | (As<Int4>(pDst.z) & ~zEnable));
+ if(dst.w) d.w = As<Float4>(As<Int4>(d.w) | (As<Int4>(pDst.w) & ~wEnable));
}
- switch(dest.type)
+ switch(dst.type)
{
- case Dst::PARAMETER_VOID:
+ case Shader::PARAMETER_VOID:
break;
- case Dst::PARAMETER_TEMP:
- if(dest.x) r.r[dest.index].x = d.x;
- if(dest.y) r.r[dest.index].y = d.y;
- if(dest.z) r.r[dest.index].z = d.z;
- if(dest.w) r.r[dest.index].w = d.w;
+ 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;
+ }
+ else
+ {
+ Int a = relativeAddress(r, 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;
+ }
break;
- case Dst::PARAMETER_ADDR:
- if(dest.x) r.a0.x = d.x;
- if(dest.y) r.a0.y = d.y;
- if(dest.z) r.a0.z = d.z;
- if(dest.w) r.a0.w = d.w;
+ case Shader::PARAMETER_ADDR:
+ if(dst.x) r.a0.x = d.x;
+ if(dst.y) r.a0.y = d.y;
+ if(dst.z) r.a0.z = d.z;
+ if(dst.w) r.a0.w = d.w;
break;
- case Dst::PARAMETER_RASTOUT:
- switch(dest.index)
+ case Shader::PARAMETER_RASTOUT:
+ switch(dst.index)
{
case 0:
- if(dest.x) r.ox[Pos] = d.x;
- if(dest.y) r.oy[Pos] = d.y;
- if(dest.z) r.oz[Pos] = d.z;
- if(dest.w) r.ow[Pos] = d.w;
+ if(dst.x) r.o[Pos].x = d.x;
+ if(dst.y) r.o[Pos].y = d.y;
+ if(dst.z) r.o[Pos].z = d.z;
+ if(dst.w) r.o[Pos].w = d.w;
break;
case 1:
- r.ox[Fog] = d.x;
+ r.o[Fog].x = d.x;
break;
case 2:
- r.oy[Pts] = d.x;
+ r.o[Pts].y = d.x;
break;
default: ASSERT(false);
}
break;
- case Dst::PARAMETER_ATTROUT:
- if(dest.x) r.ox[D0 + dest.index] = d.x;
- if(dest.y) r.oy[D0 + dest.index] = d.y;
- if(dest.z) r.oz[D0 + dest.index] = d.z;
- if(dest.w) r.ow[D0 + dest.index] = d.w;
+ case Shader::PARAMETER_ATTROUT:
+ if(dst.x) r.o[D0 + dst.index].x = d.x;
+ if(dst.y) r.o[D0 + dst.index].y = d.y;
+ if(dst.z) r.o[D0 + dst.index].z = d.z;
+ if(dst.w) r.o[D0 + dst.index].w = d.w;
break;
- case Dst::PARAMETER_TEXCRDOUT:
- // case Dst::PARAMETER_OUTPUT:
+ case Shader::PARAMETER_TEXCRDOUT:
+ // case Shader::PARAMETER_OUTPUT:
if(version < 0x0300)
{
- if(dest.x) r.ox[T0 + dest.index] = d.x;
- if(dest.y) r.oy[T0 + dest.index] = d.y;
- if(dest.z) r.oz[T0 + dest.index] = d.z;
- if(dest.w) r.ow[T0 + dest.index] = d.w;
+ if(dst.x) r.o[T0 + dst.index].x = d.x;
+ if(dst.y) r.o[T0 + dst.index].y = d.y;
+ if(dst.z) r.o[T0 + dst.index].z = d.z;
+ if(dst.w) r.o[T0 + dst.index].w = d.w;
}
else
{
- if(!dest.relative)
+ if(dst.rel.type == Shader::PARAMETER_VOID) // Not relative
{
- if(dest.x) r.ox[dest.index] = d.x;
- if(dest.y) r.oy[dest.index] = d.y;
- if(dest.z) r.oz[dest.index] = d.z;
- if(dest.w) r.ow[dest.index] = d.w;
+ if(dst.x) r.o[dst.index].x = d.x;
+ if(dst.y) r.o[dst.index].y = d.y;
+ if(dst.z) r.o[dst.index].z = d.z;
+ if(dst.w) r.o[dst.index].w = d.w;
}
- else
+ else if(dst.rel.type == Shader::PARAMETER_LOOP)
{
Int aL = r.aL[r.loopDepth];
- if(dest.x) r.ox[dest.index + aL] = d.x;
- if(dest.y) r.oy[dest.index + aL] = d.y;
- if(dest.z) r.oz[dest.index + aL] = d.z;
- if(dest.w) r.ow[dest.index + aL] = d.w;
+ if(dst.x) r.o[dst.index + aL].x = d.x;
+ if(dst.y) r.o[dst.index + aL].y = d.y;
+ if(dst.z) r.o[dst.index + aL].z = d.z;
+ if(dst.w) r.o[dst.index + aL].w = d.w;
+ }
+ else
+ {
+ Int a = relativeAddress(r, dst);
+
+ if(dst.x) r.o[dst.index + a].x = d.x;
+ if(dst.y) r.o[dst.index + a].y = d.y;
+ if(dst.z) r.o[dst.index + a].z = d.z;
+ if(dst.w) r.o[dst.index + a].w = d.w;
}
}
break;
- case Dst::PARAMETER_LABEL: break;
- case Dst::PARAMETER_PREDICATE: r.p0 = d; break;
- case Dst::PARAMETER_INPUT: break;
+ case Shader::PARAMETER_LABEL: break;
+ case Shader::PARAMETER_PREDICATE: r.p0 = d; break;
+ case Shader::PARAMETER_INPUT: break;
default:
ASSERT(false);
}
}
}
- if(returns)
+ if(currentLabel != -1)
{
Nucleus::setInsertBlock(returnBlock);
}
@@ -478,40 +524,40 @@
void VertexProgram::passThrough(Registers &r)
{
- if(vertexShader)
+ if(shader)
{
for(int i = 0; i < 12; i++)
{
- unsigned char usage = vertexShader->output[i][0].usage;
- unsigned char index = vertexShader->output[i][0].index;
+ unsigned char usage = shader->output[i][0].usage;
+ unsigned char index = shader->output[i][0].index;
switch(usage)
{
case 0xFF:
continue;
- case ShaderOperation::USAGE_PSIZE:
- r.oy[i] = r.v[i].x;
+ case Shader::USAGE_PSIZE:
+ r.o[i].y = r.v[i].x;
break;
- case ShaderOperation::USAGE_TEXCOORD:
- r.ox[i] = r.v[i].x;
- r.oy[i] = r.v[i].y;
- r.oz[i] = r.v[i].z;
- r.ow[i] = r.v[i].w;
+ case Shader::USAGE_TEXCOORD:
+ r.o[i].x = r.v[i].x;
+ r.o[i].y = r.v[i].y;
+ r.o[i].z = r.v[i].z;
+ r.o[i].w = r.v[i].w;
break;
- case ShaderOperation::USAGE_POSITION:
- r.ox[i] = r.v[i].x;
- r.oy[i] = r.v[i].y;
- r.oz[i] = r.v[i].z;
- r.ow[i] = r.v[i].w;
+ case Shader::USAGE_POSITION:
+ r.o[i].x = r.v[i].x;
+ r.o[i].y = r.v[i].y;
+ r.o[i].z = r.v[i].z;
+ r.o[i].w = r.v[i].w;
break;
- case ShaderOperation::USAGE_COLOR:
- r.ox[i] = r.v[i].x;
- r.oy[i] = r.v[i].y;
- r.oz[i] = r.v[i].z;
- r.ow[i] = r.v[i].w;
+ case Shader::USAGE_COLOR:
+ r.o[i].x = r.v[i].x;
+ r.o[i].y = r.v[i].y;
+ r.o[i].z = r.v[i].z;
+ r.o[i].w = r.v[i].w;
break;
- case ShaderOperation::USAGE_FOG:
- r.ox[i] = r.v[i].x;
+ case Shader::USAGE_FOG:
+ r.o[i].x = r.v[i].x;
break;
default:
ASSERT(false);
@@ -520,60 +566,99 @@
}
else
{
- r.ox[Pos] = r.v[PositionT].x;
- r.oy[Pos] = r.v[PositionT].y;
- r.oz[Pos] = r.v[PositionT].z;
- r.ow[Pos] = r.v[PositionT].w;
+ r.o[Pos].x = r.v[PositionT].x;
+ r.o[Pos].y = r.v[PositionT].y;
+ r.o[Pos].z = r.v[PositionT].z;
+ r.o[Pos].w = r.v[PositionT].w;
for(int i = 0; i < 2; i++)
{
- r.ox[D0 + i] = r.v[Color0 + i].x;
- r.oy[D0 + i] = r.v[Color0 + i].y;
- r.oz[D0 + i] = r.v[Color0 + i].z;
- r.ow[D0 + i] = r.v[Color0 + i].w;
+ r.o[D0 + i].x = r.v[Color0 + i].x;
+ r.o[D0 + i].y = r.v[Color0 + i].y;
+ r.o[D0 + i].z = r.v[Color0 + i].z;
+ r.o[D0 + i].w = r.v[Color0 + i].w;
}
for(int i = 0; i < 8; i++)
{
- r.ox[T0 + i] = r.v[TexCoord0 + i].x;
- r.oy[T0 + i] = r.v[TexCoord0 + i].y;
- r.oz[T0 + i] = r.v[TexCoord0 + i].z;
- r.ow[T0 + i] = r.v[TexCoord0 + i].w;
+ r.o[T0 + i].x = r.v[TexCoord0 + i].x;
+ r.o[T0 + i].y = r.v[TexCoord0 + i].y;
+ r.o[T0 + i].z = r.v[TexCoord0 + i].z;
+ r.o[T0 + i].w = r.v[TexCoord0 + i].w;
}
- r.oy[Pts] = r.v[PSize].x;
+ r.o[Pts].y = r.v[PSize].x;
}
}
- Color4f VertexProgram::reg(Registers &r, const Src &src, int offset)
+ Vector4f VertexProgram::reg(Registers &r, const Src &src, int offset)
{
int i = src.index + offset;
- Color4f reg;
+ Vector4f reg;
- if(src.type == Src::PARAMETER_CONST)
- {
- reg = readConstant(r, src, offset);
- }
-
switch(src.type)
{
- case Src::PARAMETER_TEMP: reg = r.r[i]; break;
- case Src::PARAMETER_CONST: break;
- case Src::PARAMETER_INPUT: reg = r.v[i]; break;
- case Src::PARAMETER_VOID: return r.r[0]; // Dummy
- case Src::PARAMETER_FLOATLITERAL: return r.r[0]; // Dummy
- case Src::PARAMETER_ADDR: reg = r.a0; break;
- case Src::PARAMETER_CONSTBOOL: return r.r[0]; // Dummy
- case Src::PARAMETER_CONSTINT: return r.r[0]; // Dummy
- case Src::PARAMETER_LOOP: return r.r[0]; // Dummy
- case Src::PARAMETER_PREDICATE: return r.r[0]; // Dummy
- case Src::PARAMETER_SAMPLER: return r.r[0]; // Dummy
+ case Shader::PARAMETER_TEMP:
+ if(src.rel.type == Shader::PARAMETER_VOID)
+ {
+ reg = r.r[i];
+ }
+ else
+ {
+ reg = r.r[i + relativeAddress(r, src)];
+ }
+ break;
+ case Shader::PARAMETER_CONST:
+ reg = readConstant(r, src, offset);
+ break;
+ case Shader::PARAMETER_INPUT:
+ if(src.rel.type == Shader::PARAMETER_VOID)
+ {
+ reg = r.v[i];
+ }
+ else
+ {
+ reg = r.v[i + relativeAddress(r, src)];
+ }
+ break;
+ case Shader::PARAMETER_VOID: return r.r[0]; // Dummy
+ case Shader::PARAMETER_FLOAT4LITERAL:
+ reg.x = Float4(src.value[0]);
+ reg.y = Float4(src.value[1]);
+ reg.z = Float4(src.value[2]);
+ reg.w = Float4(src.value[3]);
+ break;
+ case Shader::PARAMETER_ADDR: reg = r.a0; break;
+ case Shader::PARAMETER_CONSTBOOL: return r.r[0]; // Dummy
+ case Shader::PARAMETER_CONSTINT: return r.r[0]; // Dummy
+ case Shader::PARAMETER_LOOP: return r.r[0]; // Dummy
+ case Shader::PARAMETER_PREDICATE: return r.r[0]; // Dummy
+ case Shader::PARAMETER_SAMPLER:
+ if(src.rel.type == Shader::PARAMETER_VOID)
+ {
+ reg.x = As<Float4>(Int4(i));
+ }
+ else if(src.rel.type == Shader::PARAMETER_TEMP)
+ {
+ reg.x = As<Float4>(Int4(i) + RoundInt(r.r[src.rel.index].x));
+ }
+ return reg;
+ case Shader::PARAMETER_OUTPUT:
+ if(src.rel.type == Shader::PARAMETER_VOID)
+ {
+ reg = r.o[i];
+ }
+ else
+ {
+ reg = r.o[i + relativeAddress(r, src)];
+ }
+ break;
default:
ASSERT(false);
}
- Color4f mod;
+ Vector4f mod;
mod.x = reg[(src.swizzle >> 0) & 0x03];
mod.y = reg[(src.swizzle >> 2) & 0x03];
@@ -582,54 +667,54 @@
switch(src.modifier)
{
- case Src::MODIFIER_NONE:
+ case Shader::MODIFIER_NONE:
break;
- case Src::MODIFIER_NEGATE:
+ case Shader::MODIFIER_NEGATE:
mod.x = -mod.x;
mod.y = -mod.y;
mod.z = -mod.z;
mod.w = -mod.w;
break;
- case Src::MODIFIER_BIAS:
+ case Shader::MODIFIER_BIAS:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_BIAS_NEGATE:
+ case Shader::MODIFIER_BIAS_NEGATE:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_SIGN:
+ case Shader::MODIFIER_SIGN:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_SIGN_NEGATE:
+ case Shader::MODIFIER_SIGN_NEGATE:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_COMPLEMENT:
+ case Shader::MODIFIER_COMPLEMENT:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_X2:
+ case Shader::MODIFIER_X2:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_X2_NEGATE:
+ case Shader::MODIFIER_X2_NEGATE:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_DZ:
+ case Shader::MODIFIER_DZ:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_DW:
+ case Shader::MODIFIER_DW:
ASSERT(false); // NOTE: Unimplemented
break;
- case Src::MODIFIER_ABS:
+ case Shader::MODIFIER_ABS:
mod.x = Abs(mod.x);
mod.y = Abs(mod.y);
mod.z = Abs(mod.z);
mod.w = Abs(mod.w);
break;
- case Src::MODIFIER_ABS_NEGATE:
+ case Shader::MODIFIER_ABS_NEGATE:
mod.x = -Abs(mod.x);
mod.y = -Abs(mod.y);
mod.z = -Abs(mod.z);
mod.w = -Abs(mod.w);
break;
- case Src::MODIFIER_NOT:
+ case Shader::MODIFIER_NOT:
UNIMPLEMENTED();
break;
default:
@@ -639,32 +724,177 @@
return mod;
}
- void VertexProgram::M3X2(Registers &r, Color4f &dst, Color4f &src0, Src &src1)
+ Vector4f VertexProgram::readConstant(Registers &r, const Src &src, int offset)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
+ Vector4f c;
+
+ 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,vs.c[i]));
+
+ c.x = c.x.xxxx;
+ c.y = c.y.yyyy;
+ c.z = c.z.zzzz;
+ c.w = c.w.wwww;
+
+ if(localShaderConstants) // Constant may be known at compile time
+ {
+ for(int j = 0; j < shader->getLength(); j++)
+ {
+ const Shader::Instruction &instruction = *shader->getInstruction(j);
+
+ if(instruction.opcode == Shader::OPCODE_DEF)
+ {
+ if(instruction.dst.index == i)
+ {
+ c.x = Float4(instruction.src[0].value[0]);
+ c.y = Float4(instruction.src[0].value[1]);
+ c.z = Float4(instruction.src[0].value[2]);
+ c.w = Float4(instruction.src[0].value[3]);
+
+ break;
+ }
+ }
+ }
+ }
+ }
+ else if(src.rel.type == Shader::PARAMETER_LOOP)
+ {
+ Int loopCounter = r.aL[r.loopDepth];
+
+ c.x = c.y = c.z = c.w = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c[i]) + loopCounter * 16);
+
+ c.x = c.x.xxxx;
+ c.y = c.y.yyyy;
+ c.z = c.z.zzzz;
+ c.w = c.w.wwww;
+ }
+ else
+ {
+ if(src.rel.deterministic)
+ {
+ Int a = relativeAddress(r, src);
+
+ c.x = c.y = c.z = c.w = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c[i]) + a * 16);
+
+ c.x = c.x.xxxx;
+ c.y = c.y.yyyy;
+ c.z = c.z.zzzz;
+ c.w = c.w.wwww;
+ }
+ else
+ {
+ int component = src.rel.swizzle & 0x03;
+ Float4 a;
+
+ switch(src.rel.type)
+ {
+ case Shader::PARAMETER_ADDR: a = r.a0[component]; break;
+ case Shader::PARAMETER_TEMP: a = r.r[src.rel.index][component]; break;
+ case Shader::PARAMETER_INPUT: a = r.v[src.rel.index][component]; break;
+ case Shader::PARAMETER_OUTPUT: a = r.o[src.rel.index][component]; break;
+ case Shader::PARAMETER_CONST: a = Float4(*Pointer<Float>(r.data + OFFSET(DrawData,vs.c[src.rel.index][component]))); break;
+ default: ASSERT(false);
+ }
+
+ Int4 index = Int4(i) + RoundInt(a) * Int4(src.rel.scale);
+
+ index = Min(As<UInt4>(index), UInt4(256)); // Clamp to constant register range, c[256] = {0, 0, 0, 0}
+
+ Int index0 = Extract(index, 0);
+ Int index1 = Extract(index, 1);
+ Int index2 = Extract(index, 2);
+ Int index3 = Extract(index, 3);
+
+ c.x = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index0 * 16, 16);
+ c.y = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index1 * 16, 16);
+ c.z = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index2 * 16, 16);
+ c.w = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c) + index3 * 16, 16);
+
+ transpose4x4(c.x, c.y, c.z, c.w);
+ }
+ }
+
+ return c;
+ }
+
+ Int VertexProgram::relativeAddress(Registers &r, const Shader::Parameter &var)
+ {
+ ASSERT(var.rel.deterministic);
+
+ if(var.rel.type == Shader::PARAMETER_TEMP)
+ {
+ return RoundInt(Extract(r.r[var.rel.index].x, 0)) * var.rel.scale;
+ }
+ else if(var.rel.type == Shader::PARAMETER_INPUT)
+ {
+ return RoundInt(Extract(r.v[var.rel.index].x, 0)) * var.rel.scale;
+ }
+ else if(var.rel.type == Shader::PARAMETER_OUTPUT)
+ {
+ return RoundInt(Extract(r.o[var.rel.index].x, 0)) * var.rel.scale;
+ }
+ else if(var.rel.type == Shader::PARAMETER_CONST)
+ {
+ RValue<Float4> c = *Pointer<Float4>(r.data + OFFSET(DrawData,vs.c[var.rel.index]));
+
+ return RoundInt(Extract(c, 0)) * var.rel.scale;
+ }
+ else ASSERT(false);
+
+ return 0;
+ }
+
+ Int4 VertexProgram::enableMask(Registers &r, const Shader::Instruction *instruction)
+ {
+ Int4 enable = instruction->analysisBranch ? Int4(r.enableStack[r.enableIndex]) : Int4(0xFFFFFFFF);
+
+ if(shader->containsBreakInstruction() && !whileTest && instruction->analysisBreak)
+ {
+ enable &= r.enableBreak;
+ }
+
+ if(shader->containsContinueInstruction() && !whileTest && instruction->analysisContinue)
+ {
+ enable &= r.enableContinue;
+ }
+
+ if(shader->containsLeaveInstruction() && instruction->analysisLeave)
+ {
+ enable &= r.enableLeave;
+ }
+
+ return enable;
+ }
+
+ void VertexProgram::M3X2(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1)
+ {
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
}
- void VertexProgram::M3X3(Registers &r, Color4f &dst, Color4f &src0, Src &src1)
+ void VertexProgram::M3X3(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
dst.z = dot3(src0, row2);
}
- void VertexProgram::M3X4(Registers &r, Color4f &dst, Color4f &src0, Src &src1)
+ void VertexProgram::M3X4(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
- Color4f row3 = reg(r, src1, 3);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
+ Vector4f row3 = reg(r, src1, 3);
dst.x = dot3(src0, row0);
dst.y = dot3(src0, row1);
@@ -672,23 +902,23 @@
dst.w = dot3(src0, row3);
}
- void VertexProgram::M4X3(Registers &r, Color4f &dst, Color4f &src0, Src &src1)
+ void VertexProgram::M4X3(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1);
dst.z = dot4(src0, row2);
}
- void VertexProgram::M4X4(Registers &r, Color4f &dst, Color4f &src0, Src &src1)
+ void VertexProgram::M4X4(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1)
{
- Color4f row0 = reg(r, src1, 0);
- Color4f row1 = reg(r, src1, 1);
- Color4f row2 = reg(r, src1, 2);
- Color4f row3 = reg(r, src1, 3);
+ Vector4f row0 = reg(r, src1, 0);
+ Vector4f row1 = reg(r, src1, 1);
+ Vector4f row2 = reg(r, src1, 2);
+ Vector4f row3 = reg(r, src1, 3);
dst.x = dot4(src0, row0);
dst.y = dot4(src0, row1);
@@ -703,6 +933,7 @@
if(breakDepth == 0)
{
+ r.enableIndex = r.enableIndex - breakDepth;
Nucleus::createBr(endBlock);
}
else
@@ -710,49 +941,47 @@
r.enableBreak = r.enableBreak & ~r.enableStack[r.enableIndex];
Bool allBreak = SignMask(r.enableBreak) == 0x0;
+ r.enableIndex = r.enableIndex - breakDepth;
branch(allBreak, endBlock, deadBlock);
}
Nucleus::setInsertBlock(deadBlock);
+ r.enableIndex = r.enableIndex + breakDepth;
}
- void VertexProgram::BREAKC(Registers &r, Color4f &src0, Color4f &src1, Control control)
+ void VertexProgram::BREAKC(Registers &r, Vector4f &src0, Vector4f &src1, Control control)
{
Int4 condition;
switch(control)
{
- case Op::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
- case Op::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
- case Op::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
- case Op::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
- case Op::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
- case Op::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
+ case Shader::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
+ case Shader::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
+ case Shader::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
+ case Shader::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
default:
ASSERT(false);
}
- condition &= r.enableStack[r.enableIndex];
-
- llvm::BasicBlock *continueBlock = Nucleus::createBasicBlock();
- llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth - 1];
-
- r.enableBreak = r.enableBreak & ~condition;
- Bool allBreak = SignMask(r.enableBreak) == 0x0;
-
- branch(allBreak, endBlock, continueBlock);
- Nucleus::setInsertBlock(continueBlock);
+ BREAK(r, condition);
}
void VertexProgram::BREAKP(Registers &r, const Src &predicateRegister) // FIXME: Factor out parts common with BREAKC
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
- if(predicateRegister.modifier == Src::MODIFIER_NOT)
+ if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
+ BREAK(r, condition);
+ }
+
+ void VertexProgram::BREAK(Registers &r, Int4 &condition)
+ {
condition &= r.enableStack[r.enableIndex];
llvm::BasicBlock *continueBlock = Nucleus::createBasicBlock();
@@ -761,44 +990,61 @@
r.enableBreak = r.enableBreak & ~condition;
Bool allBreak = SignMask(r.enableBreak) == 0x0;
+ r.enableIndex = r.enableIndex - breakDepth;
branch(allBreak, endBlock, continueBlock);
+
Nucleus::setInsertBlock(continueBlock);
+ r.enableIndex = r.enableIndex + breakDepth;
}
- void VertexProgram::CALL(Registers &r, int labelIndex)
+ void VertexProgram::CONTINUE(Registers &r)
+ {
+ r.enableContinue = r.enableContinue & ~r.enableStack[r.enableIndex];
+ }
+
+ void VertexProgram::TEST()
+ {
+ whileTest = true;
+ }
+
+ void VertexProgram::CALL(Registers &r, int labelIndex, int callSiteIndex)
{
if(!labelBlock[labelIndex])
{
labelBlock[labelIndex] = Nucleus::createBasicBlock();
}
- llvm::BasicBlock *retBlock = Nucleus::createBasicBlock();
- callRetBlock.push_back(retBlock);
+ if(callRetBlock[labelIndex].size() > 1)
+ {
+ r.callStack[r.stackIndex++] = UInt(callSiteIndex);
+ }
- r.callStack[r.stackIndex++] = UInt((unsigned int)callRetBlock.size() - 1); // FIXME
+ Int4 restoreLeave = r.enableLeave;
Nucleus::createBr(labelBlock[labelIndex]);
- Nucleus::setInsertBlock(retBlock);
+ Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
+
+ r.enableLeave = restoreLeave;
}
- void VertexProgram::CALLNZ(Registers &r, int labelIndex, const Src &src)
+ void VertexProgram::CALLNZ(Registers &r, int labelIndex, int callSiteIndex, const Src &src)
{
- if(src.type == Src::PARAMETER_CONSTBOOL)
+ if(src.type == Shader::PARAMETER_CONSTBOOL)
{
- CALLNZb(r, labelIndex, src);
+ CALLNZb(r, labelIndex, callSiteIndex, src);
}
- else if(src.type == Src::PARAMETER_PREDICATE)
+ else if(src.type == Shader::PARAMETER_PREDICATE)
{
- CALLNZp(r, labelIndex, src);
+ CALLNZp(r, labelIndex, callSiteIndex, src);
}
else ASSERT(false);
}
- void VertexProgram::CALLNZb(Registers &r, int labelIndex, const Src &boolRegister)
+ void VertexProgram::CALLNZb(Registers &r, int labelIndex, int callSiteIndex, const Src &boolRegister)
{
Bool condition = (*Pointer<Byte>(r.data + OFFSET(DrawData,vs.b[boolRegister.index])) != Byte(0)); // FIXME
- if(boolRegister.modifier == Src::MODIFIER_NOT)
+ if(boolRegister.modifier == Shader::MODIFIER_NOT)
{
condition = !condition;
}
@@ -808,20 +1054,24 @@
labelBlock[labelIndex] = Nucleus::createBasicBlock();
}
- llvm::BasicBlock *retBlock = Nucleus::createBasicBlock();
- callRetBlock.push_back(retBlock);
+ if(callRetBlock[labelIndex].size() > 1)
+ {
+ r.callStack[r.stackIndex++] = UInt(callSiteIndex);
+ }
- r.callStack[r.stackIndex++] = UInt((int)callRetBlock.size() - 1); // FIXME
+ Int4 restoreLeave = r.enableLeave;
- branch(condition, labelBlock[labelIndex], retBlock);
- Nucleus::setInsertBlock(retBlock);
+ branch(condition, labelBlock[labelIndex], callRetBlock[labelIndex][callSiteIndex]);
+ Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
+
+ r.enableLeave = restoreLeave;
}
- void VertexProgram::CALLNZp(Registers &r, int labelIndex, const Src &predicateRegister)
+ void VertexProgram::CALLNZp(Registers &r, int labelIndex, int callSiteIndex, const Src &predicateRegister)
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
- if(predicateRegister.modifier == Src::MODIFIER_NOT)
+ if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
@@ -833,20 +1083,21 @@
labelBlock[labelIndex] = Nucleus::createBasicBlock();
}
- llvm::BasicBlock *retBlock = Nucleus::createBasicBlock();
- callRetBlock.push_back(retBlock);
-
- r.callStack[r.stackIndex++] = UInt((int)callRetBlock.size() - 1); // FIXME
+ if(callRetBlock[labelIndex].size() > 1)
+ {
+ r.callStack[r.stackIndex++] = UInt(callSiteIndex);
+ }
r.enableIndex++;
r.enableStack[r.enableIndex] = condition;
+ Int4 restoreLeave = r.enableLeave;
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
-
- branch(notAllFalse, labelBlock[labelIndex], retBlock);
- Nucleus::setInsertBlock(retBlock);
+ Bool notAllFalse = SignMask(condition) != 0;
+ branch(notAllFalse, labelBlock[labelIndex], callRetBlock[labelIndex][callSiteIndex]);
+ Nucleus::setInsertBlock(callRetBlock[labelIndex][callSiteIndex]);
r.enableIndex--;
+ r.enableLeave = restoreLeave;
}
void VertexProgram::ELSE(Registers &r)
@@ -859,7 +1110,7 @@
if(isConditionalIf[ifDepth])
{
Int4 condition = ~r.enableStack[r.enableIndex] & r.enableStack[r.enableIndex - 1];
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
+ Bool notAllFalse = SignMask(condition) != 0;
branch(notAllFalse, falseBlock, endBlock);
@@ -892,20 +1143,6 @@
}
}
- void VertexProgram::ENDREP(Registers &r)
- {
- loopRepDepth--;
-
- llvm::BasicBlock *testBlock = loopRepTestBlock[loopRepDepth];
- llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth];
-
- Nucleus::createBr(testBlock);
- Nucleus::setInsertBlock(endBlock);
-
- r.loopDepth--;
- r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
- }
-
void VertexProgram::ENDLOOP(Registers &r)
{
loopRepDepth--;
@@ -922,17 +1159,50 @@
r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
}
+ void VertexProgram::ENDREP(Registers &r)
+ {
+ loopRepDepth--;
+
+ llvm::BasicBlock *testBlock = loopRepTestBlock[loopRepDepth];
+ llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth];
+
+ Nucleus::createBr(testBlock);
+ Nucleus::setInsertBlock(endBlock);
+
+ r.loopDepth--;
+ r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+ }
+
+ void VertexProgram::ENDWHILE(Registers &r)
+ {
+ loopRepDepth--;
+
+ llvm::BasicBlock *testBlock = loopRepTestBlock[loopRepDepth];
+ llvm::BasicBlock *endBlock = loopRepEndBlock[loopRepDepth];
+
+ Nucleus::createBr(testBlock);
+ Nucleus::setInsertBlock(endBlock);
+
+ r.enableIndex--;
+ r.enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+ whileTest = false;
+ }
+
void VertexProgram::IF(Registers &r, const Src &src)
{
- if(src.type == Src::PARAMETER_CONSTBOOL)
+ if(src.type == Shader::PARAMETER_CONSTBOOL)
{
IFb(r, src);
}
- else if(src.type == Src::PARAMETER_PREDICATE)
+ else if(src.type == Shader::PARAMETER_PREDICATE)
{
IFp(r, src);
}
- else ASSERT(false);
+ else
+ {
+ Int4 condition = As<Int4>(reg(r, src).x);
+ IF(r, condition);
+ }
}
void VertexProgram::IFb(Registers &r, const Src &boolRegister)
@@ -941,9 +1211,9 @@
Bool condition = (*Pointer<Byte>(r.data + OFFSET(DrawData,vs.b[boolRegister.index])) != Byte(0)); // FIXME
- if(boolRegister.modifier == Src::MODIFIER_NOT)
+ if(boolRegister.modifier == Shader::MODIFIER_NOT)
{
- condition = !condition;
+ condition = !condition;
}
llvm::BasicBlock *trueBlock = Nucleus::createBasicBlock();
@@ -957,50 +1227,39 @@
ifDepth++;
}
- void VertexProgram::IFp(Registers &r, const Src &predicateRegister) // FIXME: Factor out parts common with IFC
+ void VertexProgram::IFp(Registers &r, const Src &predicateRegister)
{
Int4 condition = As<Int4>(r.p0[predicateRegister.swizzle & 0x3]);
- if(predicateRegister.modifier == Src::MODIFIER_NOT)
+ if(predicateRegister.modifier == Shader::MODIFIER_NOT)
{
condition = ~condition;
}
- condition &= r.enableStack[r.enableIndex];
-
- r.enableIndex++;
- r.enableStack[r.enableIndex] = condition;
-
- llvm::BasicBlock *trueBlock = Nucleus::createBasicBlock();
- llvm::BasicBlock *falseBlock = Nucleus::createBasicBlock();
-
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
-
- branch(notAllFalse, trueBlock, falseBlock);
-
- isConditionalIf[ifDepth] = true;
- ifFalseBlock[ifDepth] = falseBlock;
-
- ifDepth++;
- breakDepth++;
+ IF(r, condition);
}
- void VertexProgram::IFC(Registers &r, Color4f &src0, Color4f &src1, Control control)
+ void VertexProgram::IFC(Registers &r, Vector4f &src0, Vector4f &src1, Control control)
{
Int4 condition;
switch(control)
{
- case Op::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
- case Op::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
- case Op::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
- case Op::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
- case Op::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
- case Op::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
+ case Shader::CONTROL_GT: condition = CmpNLE(src0.x, src1.x); break;
+ case Shader::CONTROL_EQ: condition = CmpEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_GE: condition = CmpNLT(src0.x, src1.x); break;
+ case Shader::CONTROL_LT: condition = CmpLT(src0.x, src1.x); break;
+ case Shader::CONTROL_NE: condition = CmpNEQ(src0.x, src1.x); break;
+ case Shader::CONTROL_LE: condition = CmpLE(src0.x, src1.x); break;
default:
ASSERT(false);
}
+ IF(r, condition);
+ }
+
+ void VertexProgram::IF(Registers &r, Int4 &condition)
+ {
condition &= r.enableStack[r.enableIndex];
r.enableIndex++;
@@ -1009,7 +1268,7 @@
llvm::BasicBlock *trueBlock = Nucleus::createBasicBlock();
llvm::BasicBlock *falseBlock = Nucleus::createBasicBlock();
- Bool notAllFalse = SignMask(condition & r.enableBreak) != 0;
+ Bool notAllFalse = SignMask(condition) != 0;
branch(notAllFalse, trueBlock, falseBlock);
@@ -1022,7 +1281,13 @@
void VertexProgram::LABEL(int labelIndex)
{
+ if(!labelBlock[labelIndex])
+ {
+ labelBlock[labelIndex] = Nucleus::createBasicBlock();
+ }
+
Nucleus::setInsertBlock(labelBlock[labelIndex]);
+ currentLabel = labelIndex;
}
void VertexProgram::LOOP(Registers &r, const Src &integerRegister)
@@ -1086,27 +1351,73 @@
breakDepth = 0;
}
+ void VertexProgram::WHILE(Registers &r, const Src &temporaryRegister)
+ {
+ r.enableIndex++;
+
+ llvm::BasicBlock *loopBlock = Nucleus::createBasicBlock();
+ llvm::BasicBlock *testBlock = Nucleus::createBasicBlock();
+ llvm::BasicBlock *endBlock = Nucleus::createBasicBlock();
+
+ loopRepTestBlock[loopRepDepth] = testBlock;
+ loopRepEndBlock[loopRepDepth] = endBlock;
+
+ Int4 restoreBreak = r.enableBreak;
+ Int4 restoreContinue = r.enableContinue;
+
+ // FIXME: jump(testBlock)
+ Nucleus::createBr(testBlock);
+ Nucleus::setInsertBlock(testBlock);
+ r.enableContinue = restoreContinue;
+
+ Vector4f &src = reg(r, temporaryRegister);
+ Int4 condition = As<Int4>(src.x);
+ condition &= r.enableStack[r.enableIndex - 1];
+ r.enableStack[r.enableIndex] = condition;
+
+ Bool notAllFalse = SignMask(condition) != 0;
+ branch(notAllFalse, loopBlock, endBlock);
+
+ Nucleus::setInsertBlock(endBlock);
+ r.enableBreak = restoreBreak;
+
+ Nucleus::setInsertBlock(loopBlock);
+
+ loopRepDepth++;
+ breakDepth = 0;
+ }
+
void VertexProgram::RET(Registers &r)
{
- if(!returns)
+ if(currentLabel == -1)
{
returnBlock = Nucleus::createBasicBlock();
Nucleus::createBr(returnBlock);
-
- returns = true;
}
else
{
- // FIXME: Encapsulate
- UInt index = r.callStack[--r.stackIndex];
-
llvm::BasicBlock *unreachableBlock = Nucleus::createBasicBlock();
- llvm::Value *value = Nucleus::createLoad(index.address);
- llvm::Value *switchInst = Nucleus::createSwitch(value, unreachableBlock, (int)callRetBlock.size());
- for(unsigned int i = 0; i < callRetBlock.size(); i++)
+ if(callRetBlock[currentLabel].size() > 1) // Pop the return destination from the call stack
{
- Nucleus::addSwitchCase(switchInst, i, callRetBlock[i]);
+ // FIXME: Encapsulate
+ UInt index = r.callStack[--r.stackIndex];
+
+ llvm::Value *value = Nucleus::createLoad(index.address);
+ llvm::Value *switchInst = Nucleus::createSwitch(value, unreachableBlock, (int)callRetBlock[currentLabel].size());
+
+ for(unsigned int i = 0; i < callRetBlock[currentLabel].size(); i++)
+ {
+ Nucleus::addSwitchCase(switchInst, i, callRetBlock[currentLabel][i]);
+ }
+ }
+ else if(callRetBlock[currentLabel].size() == 1) // Jump directly to the unique return destination
+ {
+ Nucleus::createBr(callRetBlock[currentLabel][0]);
+ }
+ else // Function isn't called
+ {
+ Nucleus::createBr(unreachableBlock);
}
Nucleus::setInsertBlock(unreachableBlock);
@@ -1114,17 +1425,60 @@
}
}
- void VertexProgram::TEXLDL(Registers &r, Color4f &dst, Color4f &src0, const Src &src1)
+ void VertexProgram::LEAVE(Registers &r)
{
- Pointer<Byte> texture = r.data + OFFSET(DrawData,mipmap[16]) + src1.index * sizeof(Texture);
+ r.enableLeave = r.enableLeave & ~r.enableStack[r.enableIndex];
- Color4f tmp;
+ // FIXME: Return from function if all instances left
+ // FIXME: Use enableLeave in other control-flow constructs
+ }
- sampler[src1.index]->sampleTexture(texture, tmp, src0.x, src0.y, src0.z, src0.w, src0, src0, false, false, true);
+ void VertexProgram::TEXLDL(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
+ {
+ Vector4f tmp;
+ sampleTexture(r, tmp, src1, src0.x, src0.y, src0.z, src0.w);
dst.x = tmp[(src1.swizzle >> 0) & 0x3];
dst.y = tmp[(src1.swizzle >> 2) & 0x3];
dst.z = tmp[(src1.swizzle >> 4) & 0x3];
dst.w = tmp[(src1.swizzle >> 6) & 0x3];
}
+
+ void VertexProgram::TEX(Registers &r, Vector4f &dst, Vector4f &src0, const Src &src1)
+ {
+ Float4 lod = Float4(0.0f);
+ Vector4f tmp;
+ sampleTexture(r, tmp, src1, src0.x, src0.y, src0.z, lod);
+
+ dst.x = tmp[(src1.swizzle >> 0) & 0x3];
+ dst.y = tmp[(src1.swizzle >> 2) & 0x3];
+ dst.z = tmp[(src1.swizzle >> 4) & 0x3];
+ dst.w = tmp[(src1.swizzle >> 6) & 0x3];
+ }
+
+ void VertexProgram::sampleTexture(Registers &r, Vector4f &c, const Src &s, Float4 &u, Float4 &v, Float4 &w, Float4 &q)
+ {
+ if(s.type == Shader::PARAMETER_SAMPLER && s.rel.type == Shader::PARAMETER_VOID)
+ {
+ Pointer<Byte> texture = r.data + OFFSET(DrawData,mipmap[16]) + s.index * sizeof(Texture);
+ sampler[s.index]->sampleTexture(texture, c, u, v, w, q, r.a0, r.a0, false, false, true);
+ }
+ else
+ {
+ Int index = As<Int>(Float(reg(r, s).x.x));
+
+ for(int i = 0; i < 16; i++)
+ {
+ if(shader->usesSampler(i))
+ {
+ If(index == i)
+ {
+ Pointer<Byte> texture = r.data + OFFSET(DrawData,mipmap[16]) + i * sizeof(Texture);
+ sampler[i]->sampleTexture(texture, c, u, v, w, q, r.a0, r.a0, false, false, true);
+ // FIXME: When the sampler states are the same, we could use one sampler and just index the texture
+ }
+ }
+ }
+ }
+ }
}
diff --git a/src/Shader/VertexProgram.hpp b/src/Shader/VertexProgram.hpp
index bae97f3..7423e5b 100644
--- a/src/Shader/VertexProgram.hpp
+++ b/src/Shader/VertexProgram.hpp
@@ -1,94 +1,104 @@
-// SwiftShader Software Renderer
-//
-// Copyright(c) 2005-2011 TransGaming Inc.
-//
-// All rights reserved. No part of this software may be copied, distributed, transmitted,
-// transcribed, stored in a retrieval system, translated into any human or computer
-// language by any means, or disclosed to third parties without the explicit written
-// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
-// or implied, including but not limited to any patent rights, are granted to you.
-//
-
-#ifndef sw_VertexProgram_hpp
-#define sw_VertexProgram_hpp
-
-#include "VertexRoutine.hpp"
-#include "ShaderCore.hpp"
-
-#include "Stream.hpp"
-#include "Types.hpp"
-
-namespace sw
-{
- struct Stream;
- class VertexShader;
- class SamplerCore;
-
- class VertexProgram : public VertexRoutine, public ShaderCore
- {
- public:
- VertexProgram(const VertexProcessor::State &state, const VertexShader *vertexShader);
-
- virtual ~VertexProgram();
-
- private:
- typedef Shader::Instruction::DestinationParameter Dst;
- typedef Shader::Instruction::SourceParameter Src;
- typedef Shader::Instruction::Operation Op;
- typedef Shader::Instruction::Operation::Control Control;
- typedef Shader::Instruction::Operation::Usage Usage;
-
- Color4f readConstant(Registers &r, const Src &src, int offset = 0);
- void pipeline(Registers &r);
- void shader(Registers &r);
- void passThrough(Registers &r);
-
- Color4f reg(Registers &r, const Src &src, int offset = 0);
-
- void M3X2(Registers &r, Color4f &dst, Color4f &src0, Src &src1);
- void M3X3(Registers &r, Color4f &dst, Color4f &src0, Src &src1);
- void M3X4(Registers &r, Color4f &dst, Color4f &src0, Src &src1);
- void M4X3(Registers &r, Color4f &dst, Color4f &src0, Src &src1);
- void M4X4(Registers &r, Color4f &dst, Color4f &src0, Src &src1);
- void BREAK(Registers &r);
- void BREAKC(Registers &r, Color4f &src0, Color4f &src1, Control);
- void BREAKP(Registers &r, const Src &predicateRegister);
- void CALL(Registers &r, int labelIndex);
- void CALLNZ(Registers &r, int labelIndex, const Src &src);
- void CALLNZb(Registers &r, int labelIndex, const Src &boolRegister);
- void CALLNZp(Registers &r, int labelIndex, const Src &predicateRegister);
- void ELSE(Registers &r);
- void ENDIF(Registers &r);
- void ENDLOOP(Registers &r);
- void ENDREP(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, Color4f &src0, Color4f &src1, Control);
- void LABEL(int labelIndex);
- void LOOP(Registers &r, const Src &integerRegister);
- void REP(Registers &r, const Src &integerRegister);
- void RET(Registers &r);
- void TEXLDL(Registers &r, Color4f &dst, Color4f &src, const Src&);
-
- SamplerCore *sampler[4];
-
- bool returns;
- int ifDepth;
- int loopRepDepth;
- int breakDepth;
-
- // 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;
- llvm::BasicBlock *returnBlock;
- bool isConditionalIf[24 + 24];
-
- const VertexShader *const vertexShader;
- };
-}
-
-#endif // sw_VertexProgram_hpp
+// SwiftShader Software Renderer
+//
+// Copyright(c) 2005-2012 TransGaming Inc.
+//
+// All rights reserved. No part of this software may be copied, distributed, transmitted,
+// transcribed, stored in a retrieval system, translated into any human or computer
+// language by any means, or disclosed to third parties without the explicit written
+// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
+// or implied, including but not limited to any patent rights, are granted to you.
+//
+
+#ifndef sw_VertexProgram_hpp
+#define sw_VertexProgram_hpp
+
+#include "VertexRoutine.hpp"
+#include "ShaderCore.hpp"
+
+#include "Stream.hpp"
+#include "Types.hpp"
+
+namespace sw
+{
+ struct Stream;
+ class VertexShader;
+ class SamplerCore;
+
+ class VertexProgram : public VertexRoutine, public ShaderCore
+ {
+ public:
+ VertexProgram(const VertexProcessor::State &state, const VertexShader *vertexShader);
+
+ virtual ~VertexProgram();
+
+ private:
+ typedef Shader::DestinationParameter Dst;
+ typedef Shader::SourceParameter Src;
+ typedef Shader::Control Control;
+ typedef Shader::Usage Usage;
+
+ void pipeline(Registers &r);
+ void program(Registers &r);
+ void passThrough(Registers &r);
+
+ Vector4f reg(Registers &r, const Src &src, int offset = 0);
+ Vector4f readConstant(Registers &r, const Src &src, int offset = 0);
+ Int relativeAddress(Registers &r, const Shader::Parameter &var);
+ Int4 enableMask(Registers &r, const Shader::Instruction *instruction);
+
+ void M3X2(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1);
+ void M3X3(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1);
+ void M3X4(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1);
+ void M4X3(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1);
+ void M4X4(Registers &r, Vector4f &dst, Vector4f &src0, Src &src1);
+ 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);
+ void TEXLDL(Registers &r, Vector4f &dst, Vector4f &src, const Src&);
+ void TEX(Registers &r, Vector4f &dst, Vector4f &src, const Src&);
+
+ void sampleTexture(Registers &r, Vector4f &c, const Src &s, Float4 &u, Float4 &v, Float4 &w, Float4 &q);
+
+ SamplerCore *sampler[4];
+
+ 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;
+ bool isConditionalIf[24 + 24];
+ };
+}
+
+#endif // sw_VertexProgram_hpp
diff --git a/src/Shader/VertexRoutine.cpp b/src/Shader/VertexRoutine.cpp
index f28a741..1c2be07 100644
--- a/src/Shader/VertexRoutine.cpp
+++ b/src/Shader/VertexRoutine.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -20,7 +20,10 @@
namespace sw
{
- VertexRoutine::VertexRoutine(const VertexProcessor::State &state) : state(state)
+ extern bool halfIntegerCoordinates; // Pixel centers are not at integer coordinates
+ extern bool symmetricNormalizedDepth; // [-1, 1] instead of [0, 1]
+
+ VertexRoutine::VertexRoutine(const VertexProcessor::State &state, const VertexShader *shader) : state(state), shader(shader)
{
routine = 0;
}
@@ -46,7 +49,7 @@
UInt count = *Pointer<UInt>(task+ OFFSET(VertexTask,count));
- Registers r;
+ Registers r(shader);
r.data = data;
r.constants = *Pointer<Pointer<Byte>>(data + OFFSET(DrawData,constants));
@@ -82,7 +85,7 @@
Return();
}
- routine = function(L"VertexRoutine_%0.16llX", state.shaderHash);
+ routine = function(L"VertexRoutine_%0.8X", state.shaderID);
}
Routine *VertexRoutine::getRoutine()
@@ -108,41 +111,41 @@
// Backtransform
if(state.preTransformed)
{
- Float4 rhw = Float4(1.0f, 1.0f, 1.0f, 1.0f) / r.ow[pos];
+ Float4 rhw = Float4(1.0f) / r.o[pos].w;
- Float4 W = *Pointer<Float4>(r.data + OFFSET(DrawData,WWWWx16)) * Float4(1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f);
- Float4 H = *Pointer<Float4>(r.data + OFFSET(DrawData,HHHHx16)) * Float4(1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f);
- Float4 L = *Pointer<Float4>(r.data + OFFSET(DrawData,LLLLx16)) * Float4(1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f);
- Float4 T = *Pointer<Float4>(r.data + OFFSET(DrawData,TTTTx16)) * Float4(1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f, 1.0f / 16.0f);
+ Float4 W = *Pointer<Float4>(r.data + OFFSET(DrawData,Wx16)) * Float4(1.0f / 16.0f);
+ Float4 H = *Pointer<Float4>(r.data + OFFSET(DrawData,Hx16)) * Float4(1.0f / 16.0f);
+ Float4 L = *Pointer<Float4>(r.data + OFFSET(DrawData,X0x16)) * Float4(1.0f / 16.0f);
+ Float4 T = *Pointer<Float4>(r.data + OFFSET(DrawData,Y0x16)) * Float4(1.0f / 16.0f);
- r.ox[pos] = (r.ox[pos] - L) / W * rhw;
- r.oy[pos] = (r.oy[pos] - T) / H * rhw;
- r.oz[pos] = r.oz[pos] * rhw;
- r.ow[pos] = rhw;
+ r.o[pos].x = (r.o[pos].x - L) / W * rhw;
+ r.o[pos].y = (r.o[pos].y - T) / H * rhw;
+ r.o[pos].z = r.o[pos].z * rhw;
+ r.o[pos].w = rhw;
}
if(state.superSampling)
{
- r.ox[pos] = r.ox[pos] + *Pointer<Float4>(r.data + OFFSET(DrawData,XXXX)) * r.ow[pos];
- r.oy[pos] = r.oy[pos] + *Pointer<Float4>(r.data + OFFSET(DrawData,YYYY)) * r.ow[pos];
+ r.o[pos].x = r.o[pos].x + *Pointer<Float4>(r.data + OFFSET(DrawData,XXXX)) * r.o[pos].w;
+ r.o[pos].y = r.o[pos].y + *Pointer<Float4>(r.data + OFFSET(DrawData,YYYY)) * r.o[pos].w;
}
- Float4 clipX = r.ox[pos];
- Float4 clipY = r.oy[pos];
+ Float4 clipX = r.o[pos].x;
+ Float4 clipY = r.o[pos].y;
if(state.multiSampling) // Clip at pixel edges instead of pixel centers
{
- clipX += *Pointer<Float4>(r.data + OFFSET(DrawData,offX)) * r.ow[pos];
- clipY += *Pointer<Float4>(r.data + OFFSET(DrawData,offY)) * r.ow[pos];
+ clipX += *Pointer<Float4>(r.data + OFFSET(DrawData,halfPixelX)) * r.o[pos].w;
+ clipY += *Pointer<Float4>(r.data + OFFSET(DrawData,halfPixelY)) * r.o[pos].w;
}
- Int4 maxX = CmpLT(r.ow[pos], clipX);
- Int4 maxY = CmpLT(r.ow[pos], clipY);
- Int4 maxZ = CmpLT(r.ow[pos], r.oz[pos]);
+ Int4 maxX = CmpLT(r.o[pos].w, clipX);
+ Int4 maxY = CmpLT(r.o[pos].w, clipY);
+ Int4 maxZ = CmpLT(r.o[pos].w, r.o[pos].z);
- Int4 minX = CmpNLE(-r.ow[pos], clipX);
- Int4 minY = CmpNLE(-r.ow[pos], clipY);
- Int4 minZ = CmpNLE(Float4(0.0f, 0.0f, 0.0f, 0.0f), r.oz[pos]);
+ Int4 minX = CmpNLE(-r.o[pos].w, clipX);
+ Int4 minY = CmpNLE(-r.o[pos].w, clipY);
+ Int4 minZ = CmpNLE(Float4(0.0f), r.o[pos].z);
Int flags;
@@ -159,9 +162,9 @@
flags = SignMask(minZ);
r.clipFlags |= *Pointer<Int>(r.constants + OFFSET(Constants,minZ) + flags * 4);
- Int4 finiteX = CmpLE(Abs(r.ox[pos]), *Pointer<Float4>(r.constants + OFFSET(Constants,maxPos)));
- Int4 finiteY = CmpLE(Abs(r.oy[pos]), *Pointer<Float4>(r.constants + OFFSET(Constants,maxPos)));
- Int4 finiteZ = CmpLE(Abs(r.oz[pos]), *Pointer<Float4>(r.constants + OFFSET(Constants,maxPos)));
+ Int4 finiteX = CmpLE(Abs(r.o[pos].x), *Pointer<Float4>(r.constants + OFFSET(Constants,maxPos)));
+ Int4 finiteY = CmpLE(Abs(r.o[pos].y), *Pointer<Float4>(r.constants + OFFSET(Constants,maxPos)));
+ Int4 finiteZ = CmpLE(Abs(r.o[pos].z), *Pointer<Float4>(r.constants + OFFSET(Constants,maxPos)));
flags = SignMask(finiteX & finiteY & finiteZ);
r.clipFlags |= *Pointer<Int>(r.constants + OFFSET(Constants,fini) + flags * 4);
@@ -172,11 +175,11 @@
}
}
- Color4f VertexRoutine::readStream(Registers &r, Pointer<Byte> &buffer, UInt &stride, const Stream &stream, const UInt &index)
+ Vector4f VertexRoutine::readStream(Registers &r, Pointer<Byte> &buffer, UInt &stride, const Stream &stream, const UInt &index)
{
const bool texldl = state.shaderContainsTexldl;
- Color4f v;
+ Vector4f v;
Pointer<Byte> source0 = buffer + index * stride;
Pointer<Byte> source1 = source0 + (!texldl ? stride : 0);
@@ -341,8 +344,8 @@
transpose4x3(v.x, v.y, v.z, v.w);
- v.y *= Float4(1.0f / 0x00000400, 1.0f / 0x00000400, 1.0f / 0x00000400, 1.0f / 0x00000400);
- v.z *= Float4(1.0f / 0x00100000, 1.0f / 0x00100000, 1.0f / 0x00100000, 1.0f / 0x00100000);
+ v.y *= Float4(1.0f / 0x00000400);
+ v.z *= Float4(1.0f / 0x00100000);
}
break;
case STREAMTYPE_DEC3N:
@@ -390,9 +393,9 @@
transpose4x3(v.x, v.y, v.z, v.w);
- v.x *= Float4(1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f);
- v.y *= Float4(1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f);
- v.z *= Float4(1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f, 1.0f / 0x00400000 / 511.0f);
+ v.x *= Float4(1.0f / 0x00400000 / 511.0f);
+ v.y *= Float4(1.0f / 0x00400000 / 511.0f);
+ v.z *= Float4(1.0f / 0x00400000 / 511.0f);
}
break;
case STREAMTYPE_FIXED:
@@ -472,10 +475,10 @@
ASSERT(false);
}
- if(stream.count < 1) v.x = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- if(stream.count < 2) v.y = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- if(stream.count < 3) v.z = Float4(0.0f, 0.0f, 0.0f, 0.0f);
- if(stream.count < 4) v.w = Float4(1.0f, 1.0f, 1.0f, 1.0f);
+ if(stream.count < 1) v.x = Float4(0.0f);
+ if(stream.count < 2) v.y = Float4(0.0f);
+ if(stream.count < 3) v.z = Float4(0.0f);
+ if(stream.count < 4) v.w = Float4(1.0f);
return v;
}
@@ -484,55 +487,53 @@
{
int pos = state.positionRegister;
- if(state.postTransform && !state.preTransformed)
+ if(halfIntegerCoordinates)
{
- Float4 posScale = *Pointer<Float4>(r.data + OFFSET(DrawData,posScale)); // FIXME: Unpack
+ r.o[pos].x = r.o[pos].x - *Pointer<Float4>(r.data + OFFSET(DrawData,halfPixelX)) * r.o[pos].w;
+ r.o[pos].y = r.o[pos].y - *Pointer<Float4>(r.data + OFFSET(DrawData,halfPixelY)) * r.o[pos].w;
+ }
- r.ox[pos] = r.ox[pos] * posScale.x;
- r.oy[pos] = r.oy[pos] * posScale.y;
-
- Float4 posOffset = *Pointer<Float4>(r.data + OFFSET(DrawData,posOffset)); // FIXME: Unpack
-
- r.ox[pos] = r.ox[pos] + r.ow[pos] * posOffset.x;
- r.oy[pos] = r.oy[pos] + r.ow[pos] * posOffset.y;
+ if(symmetricNormalizedDepth)
+ {
+ r.o[pos].z = (r.o[pos].z + r.o[pos].w) * Float4(0.5f);
}
}
void VertexRoutine::writeCache(Pointer<Byte> &cacheLine, Registers &r)
{
- Color4f v;
+ Vector4f v;
for(int i = 0; i < 12; i++)
{
if(state.output[i].write)
{
- v.x = r.ox[i];
- v.y = r.oy[i];
- v.z = r.oz[i];
- v.w = r.ow[i];
+ v.x = r.o[i].x;
+ v.y = r.o[i].y;
+ v.z = r.o[i].z;
+ v.w = r.o[i].w;
if(state.output[i].xClamp)
{
- v.x = Max(v.x, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- v.x = Min(v.x, Float4(1.0f, 1.0f, 1.0f, 1.0f));
+ v.x = Max(v.x, Float4(0.0f));
+ v.x = Min(v.x, Float4(1.0f));
}
if(state.output[i].yClamp)
{
- v.y = Max(v.y, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- v.y = Min(v.y, Float4(1.0f, 1.0f, 1.0f, 1.0f));
+ v.y = Max(v.y, Float4(0.0f));
+ v.y = Min(v.y, Float4(1.0f));
}
if(state.output[i].zClamp)
{
- v.z = Max(v.z, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- v.z = Min(v.z, Float4(1.0f, 1.0f, 1.0f, 1.0f));
+ v.z = Max(v.z, Float4(0.0f));
+ v.z = Min(v.z, Float4(1.0f));
}
if(state.output[i].wClamp)
{
- v.w = Max(v.w, Float4(0.0f, 0.0f, 0.0f, 0.0f));
- v.w = Min(v.w, Float4(1.0f, 1.0f, 1.0f, 1.0f));
+ v.w = Max(v.w, Float4(0.0f));
+ v.w = Min(v.w, Float4(1.0f));
}
if(state.output[i].write == 0x01)
@@ -568,16 +569,16 @@
int pos = state.positionRegister;
- v.x = r.ox[pos];
- v.y = r.oy[pos];
- v.z = r.oz[pos];
- v.w = r.ow[pos];
+ v.x = r.o[pos].x;
+ v.y = r.o[pos].y;
+ v.z = r.o[pos].z;
+ v.w = r.o[pos].w;
- Float4 w = As<Float4>(As<Int4>(v.w) | (As<Int4>(CmpEQ(v.w, Float4(0, 0, 0, 0))) & As<Int4>(Float4(1, 1, 1, 1))));
+ Float4 w = As<Float4>(As<Int4>(v.w) | (As<Int4>(CmpEQ(v.w, Float4(0.0f))) & As<Int4>(Float4(1.0f))));
Float4 rhw = Float4(1.0f) / w;
- v.x = As<Float4>(RoundInt(*Pointer<Float4>(r.data + OFFSET(DrawData,LLLLx16)) + v.x * rhw * *Pointer<Float4>(r.data + OFFSET(DrawData,WWWWx16))));
- v.y = As<Float4>(RoundInt(*Pointer<Float4>(r.data + OFFSET(DrawData,TTTTx16)) + v.y * rhw * *Pointer<Float4>(r.data + OFFSET(DrawData,HHHHx16))));
+ v.x = As<Float4>(RoundInt(*Pointer<Float4>(r.data + OFFSET(DrawData,X0x16)) + v.x * rhw * *Pointer<Float4>(r.data + OFFSET(DrawData,Wx16))));
+ v.y = As<Float4>(RoundInt(*Pointer<Float4>(r.data + OFFSET(DrawData,Y0x16)) + v.y * rhw * *Pointer<Float4>(r.data + OFFSET(DrawData,Hx16))));
v.z = v.z * rhw;
v.w = rhw;
diff --git a/src/Shader/VertexRoutine.hpp b/src/Shader/VertexRoutine.hpp
index 9c21662..72323f2 100644
--- a/src/Shader/VertexRoutine.hpp
+++ b/src/Shader/VertexRoutine.hpp
@@ -1,88 +1,105 @@
-// SwiftShader Software Renderer
-//
-// Copyright(c) 2005-2011 TransGaming Inc.
-//
-// All rights reserved. No part of this software may be copied, distributed, transmitted,
-// transcribed, stored in a retrieval system, translated into any human or computer
-// language by any means, or disclosed to third parties without the explicit written
-// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
-// or implied, including but not limited to any patent rights, are granted to you.
-//
-
-#ifndef sw_VertexRoutine_hpp
-#define sw_VertexRoutine_hpp
-
-#include "Renderer/Color.hpp"
-#include "Renderer/VertexProcessor.hpp"
-#include "Reactor/Reactor.hpp"
-
-namespace sw
-{
- class VertexRoutine
- {
- protected:
- struct Registers
- {
- Registers() : callStack(4), aL(4), increment(4), iteration(4), enableStack(1 + 24), ox(12), oy(12), oz(12), ow(12)
- {
- loopDepth = -1;
- enableStack[0] = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
- enableBreak = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
- }
-
- Pointer<Byte> data;
- Pointer<Byte> constants;
-
- Array<Float4> ox;
- Array<Float4> oy;
- Array<Float4> oz;
- Array<Float4> ow;
-
- Int clipFlags;
-
- Color4f v[16];
- Color4f r[32];
- Color4f a0;
- Array<Int> aL;
- Color4f p0;
-
- Array<Int> increment;
- Array<Int> iteration;
-
- Int loopDepth;
- Int stackIndex; // FIXME: Inc/decrement callStack
- Array<UInt> callStack;
-
- Int enableIndex;
- Array<Int4> enableStack;
- Int4 enableBreak;
- };
-
- public:
- VertexRoutine(const VertexProcessor::State &state);
-
- virtual ~VertexRoutine();
-
- void generate();
- Routine *getRoutine();
-
- protected:
- const VertexProcessor::State &state;
-
- private:
- virtual void pipeline(Registers &r) = 0;
-
- typedef VertexProcessor::State::Input Stream;
-
- Color4f readStream(Registers &r, Pointer<Byte> &buffer, UInt &stride, const Stream &stream, const UInt &index);
- void readInput(Registers &r, UInt &index);
- void computeClipFlags(Registers &r);
- void postTransform(Registers &r);
- void writeCache(Pointer<Byte> &cacheLine, Registers &r);
- void writeVertex(Pointer<Byte> &vertex, Pointer<Byte> &cacheLine);
-
- Routine *routine;
- };
-}
-
-#endif // sw_VertexRoutine_hpp
+// SwiftShader Software Renderer
+//
+// Copyright(c) 2005-2012 TransGaming Inc.
+//
+// All rights reserved. No part of this software may be copied, distributed, transmitted,
+// transcribed, stored in a retrieval system, translated into any human or computer
+// language by any means, or disclosed to third parties without the explicit written
+// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
+// or implied, including but not limited to any patent rights, are granted to you.
+//
+
+#ifndef sw_VertexRoutine_hpp
+#define sw_VertexRoutine_hpp
+
+#include "Renderer/Color.hpp"
+#include "Renderer/VertexProcessor.hpp"
+#include "ShaderCore.hpp"
+#include "VertexShader.hpp"
+
+namespace sw
+{
+ class VertexRoutine
+ {
+ protected:
+ struct Registers
+ {
+ Registers(const VertexShader *shader) :
+ r(shader && shader->dynamicallyIndexedTemporaries),
+ v(shader && shader->dynamicallyIndexedInput),
+ o(shader && shader->dynamicallyIndexedOutput)
+ {
+ 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);
+ }
+
+ if(shader && shader->containsLeaveInstruction())
+ {
+ enableLeave = Int4(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+ }
+ }
+
+ Pointer<Byte> data;
+ Pointer<Byte> constants;
+
+ Int clipFlags;
+
+ RegisterArray<16> v;
+ RegisterArray<4096> r;
+ RegisterArray<12> o;
+ Vector4f a0;
+ Array<Int, 4> aL;
+ Vector4f p0;
+
+ Array<Int, 4> increment;
+ Array<Int, 4> iteration;
+
+ Int loopDepth;
+ Int stackIndex; // FIXME: Inc/decrement callStack
+ Array<UInt, 4> callStack;
+
+ Int enableIndex;
+ Array<Int4, 1 + 24> enableStack;
+ Int4 enableBreak;
+ Int4 enableContinue;
+ Int4 enableLeave;
+ };
+
+ public:
+ VertexRoutine(const VertexProcessor::State &state, const VertexShader *shader);
+
+ virtual ~VertexRoutine();
+
+ void generate();
+ Routine *getRoutine();
+
+ protected:
+ const VertexProcessor::State &state;
+ const VertexShader *const shader;
+
+ private:
+ virtual void pipeline(Registers &r) = 0;
+
+ typedef VertexProcessor::State::Input Stream;
+
+ Vector4f readStream(Registers &r, Pointer<Byte> &buffer, UInt &stride, const Stream &stream, const UInt &index);
+ void readInput(Registers &r, UInt &index);
+ void computeClipFlags(Registers &r);
+ void postTransform(Registers &r);
+ void writeCache(Pointer<Byte> &cacheLine, Registers &r);
+ void writeVertex(Pointer<Byte> &vertex, Pointer<Byte> &cacheLine);
+
+ Routine *routine;
+ };
+}
+
+#endif // sw_VertexRoutine_hpp
diff --git a/src/Shader/VertexShader.cpp b/src/Shader/VertexShader.cpp
index c958820..d5b1171 100644
--- a/src/Shader/VertexShader.cpp
+++ b/src/Shader/VertexShader.cpp
@@ -1,6 +1,6 @@
// SwiftShader Software Renderer
//
-// Copyright(c) 2005-2011 TransGaming Inc.
+// Copyright(c) 2005-2012 TransGaming Inc.
//
// All rights reserved. No part of this software may be copied, distributed, transmitted,
// transcribed, stored in a retrieval system, translated into any human or computer
@@ -16,50 +16,53 @@
namespace sw
{
- VertexShader::VertexShader(const unsigned long *token) : Shader(token)
+ VertexShader::VertexShader(const VertexShader *vs) : Shader()
+ {
+ version = 0x0300;
+ positionRegister = Pos;
+ pointSizeRegister = -1; // No vertex point size
+
+ for(int i = 0; i < 16; i++)
+ {
+ input[i] = Semantic(-1, -1);
+ }
+
+ if(vs) // Make a copy
+ {
+ for(int i = 0; i < vs->getLength(); i++)
+ {
+ append(new sw::Shader::Instruction(*vs->getInstruction(i)));
+ }
+
+ memcpy(output, vs->output, sizeof(output));
+ memcpy(input, vs->input, sizeof(input));
+ positionRegister = vs->positionRegister;
+ pointSizeRegister = vs->pointSizeRegister;
+ usedSamplers = vs->usedSamplers;
+
+ analyze();
+ }
+ }
+
+ VertexShader::VertexShader(const unsigned long *token) : Shader()
{
parse(token);
+
+ positionRegister = Pos;
+ pointSizeRegister = -1; // No vertex point size
+
+ for(int i = 0; i < 16; i++)
+ {
+ input[i] = Semantic(-1, -1);
+ }
+
+ analyze();
}
VertexShader::~VertexShader()
{
}
- void VertexShader::parse(const unsigned long *token)
- {
- minorVersion = (unsigned char)(token[0] & 0x000000FF);
- majorVersion = (unsigned char)((token[0] & 0x0000FF00) >> 8);
- shaderType = (ShaderType)((token[0] & 0xFFFF0000) >> 16);
-
- length = validate(token);
- ASSERT(length != 0);
-
- instruction = new Shader::Instruction*[length];
-
- for(int i = 0; i < length; i++)
- {
- while((*token & 0x0000FFFF) == 0x0000FFFE) // Comment token
- {
- int length = (*token & 0x7FFF0000) >> 16;
-
- token += length + 1;
- }
-
- int tokenCount = size(*token);
-
- instruction[i] = new Instruction(token, tokenCount, majorVersion);
-
- token += 1 + tokenCount;
- }
-
- analyzeInput();
- analyzeOutput();
- analyzeDirtyConstants();
- analyzeTexldl();
- analyzeDynamicBranching();
- analyzeSamplers();
- }
-
int VertexShader::validate(const unsigned long *const token)
{
if(!token)
@@ -89,36 +92,36 @@
}
else
{
- ShaderOpcode opcode = (ShaderOpcode)(token[i] & 0x0000FFFF);
+ Shader::Opcode opcode = (Shader::Opcode)(token[i] & 0x0000FFFF);
switch(opcode)
{
- case ShaderOperation::OPCODE_TEXCOORD:
- case ShaderOperation::OPCODE_TEXKILL:
- case ShaderOperation::OPCODE_TEX:
- case ShaderOperation::OPCODE_TEXBEM:
- case ShaderOperation::OPCODE_TEXBEML:
- case ShaderOperation::OPCODE_TEXREG2AR:
- case ShaderOperation::OPCODE_TEXREG2GB:
- case ShaderOperation::OPCODE_TEXM3X2PAD:
- case ShaderOperation::OPCODE_TEXM3X2TEX:
- case ShaderOperation::OPCODE_TEXM3X3PAD:
- case ShaderOperation::OPCODE_TEXM3X3TEX:
- case ShaderOperation::OPCODE_RESERVED0:
- case ShaderOperation::OPCODE_TEXM3X3SPEC:
- case ShaderOperation::OPCODE_TEXM3X3VSPEC:
- case ShaderOperation::OPCODE_TEXREG2RGB:
- case ShaderOperation::OPCODE_TEXDP3TEX:
- case ShaderOperation::OPCODE_TEXM3X2DEPTH:
- case ShaderOperation::OPCODE_TEXDP3:
- case ShaderOperation::OPCODE_TEXM3X3:
- case ShaderOperation::OPCODE_TEXDEPTH:
- case ShaderOperation::OPCODE_CMP:
- case ShaderOperation::OPCODE_BEM:
- case ShaderOperation::OPCODE_DP2ADD:
- case ShaderOperation::OPCODE_DSX:
- case ShaderOperation::OPCODE_DSY:
- case ShaderOperation::OPCODE_TEXLDD:
+ case Shader::OPCODE_TEXCOORD:
+ case Shader::OPCODE_TEXKILL:
+ case Shader::OPCODE_TEX:
+ case Shader::OPCODE_TEXBEM:
+ case Shader::OPCODE_TEXBEML:
+ case Shader::OPCODE_TEXREG2AR:
+ case Shader::OPCODE_TEXREG2GB:
+ case Shader::OPCODE_TEXM3X2PAD:
+ case Shader::OPCODE_TEXM3X2TEX:
+ case Shader::OPCODE_TEXM3X3PAD:
+ case Shader::OPCODE_TEXM3X3TEX:
+ case Shader::OPCODE_RESERVED0:
+ case Shader::OPCODE_TEXM3X3SPEC:
+ case Shader::OPCODE_TEXM3X3VSPEC:
+ case Shader::OPCODE_TEXREG2RGB:
+ case Shader::OPCODE_TEXDP3TEX:
+ case Shader::OPCODE_TEXM3X2DEPTH:
+ case Shader::OPCODE_TEXDP3:
+ case Shader::OPCODE_TEXM3X3:
+ case Shader::OPCODE_TEXDEPTH:
+ case Shader::OPCODE_CMP0:
+ case Shader::OPCODE_BEM:
+ case Shader::OPCODE_DP2ADD:
+ case Shader::OPCODE_DFDX:
+ case Shader::OPCODE_DFDY:
+ case Shader::OPCODE_TEXLDD:
return 0; // Unsupported operation
default:
instructionCount++;
@@ -137,81 +140,85 @@
return texldl;
}
+ void VertexShader::analyze()
+ {
+ analyzeInput();
+ analyzeOutput();
+ analyzeDirtyConstants();
+ analyzeTexldl();
+ analyzeDynamicBranching();
+ analyzeSamplers();
+ analyzeCallSites();
+ analyzeDynamicIndexing();
+ }
+
void VertexShader::analyzeInput()
{
- for(int i = 0; i < 16; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- input[i] = Semantic(-1, -1);
- }
-
- for(int i = 0; i < length; i++)
- {
- if(instruction[i]->getOpcode() == ShaderOperation::OPCODE_DCL &&
- instruction[i]->getDestinationParameter().type == ShaderParameter::PARAMETER_INPUT)
+ if(instruction[i]->opcode == Shader::OPCODE_DCL &&
+ instruction[i]->dst.type == Shader::PARAMETER_INPUT)
{
- int index = instruction[i]->getDestinationParameter().index;
+ int index = instruction[i]->dst.index;
- input[index] = Semantic(instruction[i]->getUsage(), instruction[i]->getUsageIndex());
+ input[index] = Semantic(instruction[i]->usage, instruction[i]->usageIndex);
}
}
}
void VertexShader::analyzeOutput()
{
- positionRegister = Pos;
- pointSizeRegister = -1; // No vertex point size
-
if(version < 0x0300)
{
- output[Pos][0] = Semantic(ShaderOperation::USAGE_POSITION, 0);
- output[Pos][1] = Semantic(ShaderOperation::USAGE_POSITION, 0);
- output[Pos][2] = Semantic(ShaderOperation::USAGE_POSITION, 0);
- output[Pos][3] = Semantic(ShaderOperation::USAGE_POSITION, 0);
+ output[Pos][0] = Semantic(Shader::USAGE_POSITION, 0);
+ output[Pos][1] = Semantic(Shader::USAGE_POSITION, 0);
+ output[Pos][2] = Semantic(Shader::USAGE_POSITION, 0);
+ output[Pos][3] = Semantic(Shader::USAGE_POSITION, 0);
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- const Instruction::DestinationParameter &dst = instruction[i]->getDestinationParameter();
+ const DestinationParameter &dst = instruction[i]->dst;
switch(dst.type)
{
- case ShaderParameter::PARAMETER_RASTOUT:
+ case Shader::PARAMETER_RASTOUT:
switch(dst.index)
{
case 0:
// Position already assumed written
break;
case 1:
- output[Fog][0] = Semantic(ShaderOperation::USAGE_FOG, 0);
+ output[Fog][0] = Semantic(Shader::USAGE_FOG, 0);
break;
case 2:
- output[Pts][1] = Semantic(ShaderOperation::USAGE_PSIZE, 0);
+ output[Pts][1] = Semantic(Shader::USAGE_PSIZE, 0);
pointSizeRegister = Pts;
break;
default: ASSERT(false);
}
break;
- case ShaderParameter::PARAMETER_ATTROUT:
+ case Shader::PARAMETER_ATTROUT:
if(dst.index == 0)
{
- if(dst.x) output[D0][0] = Semantic(ShaderOperation::USAGE_COLOR, 0);
- if(dst.y) output[D0][1] = Semantic(ShaderOperation::USAGE_COLOR, 0);
- if(dst.z) output[D0][2] = Semantic(ShaderOperation::USAGE_COLOR, 0);
- if(dst.w) output[D0][3] = Semantic(ShaderOperation::USAGE_COLOR, 0);
+ if(dst.x) output[D0][0] = Semantic(Shader::USAGE_COLOR, 0);
+ if(dst.y) output[D0][1] = Semantic(Shader::USAGE_COLOR, 0);
+ if(dst.z) output[D0][2] = Semantic(Shader::USAGE_COLOR, 0);
+ if(dst.w) output[D0][3] = Semantic(Shader::USAGE_COLOR, 0);
}
else if(dst.index == 1)
{
- if(dst.x) output[D1][0] = Semantic(ShaderOperation::USAGE_COLOR, 1);
- if(dst.y) output[D1][1] = Semantic(ShaderOperation::USAGE_COLOR, 1);
- if(dst.z) output[D1][2] = Semantic(ShaderOperation::USAGE_COLOR, 1);
- if(dst.w) output[D1][3] = Semantic(ShaderOperation::USAGE_COLOR, 1);
+ if(dst.x) output[D1][0] = Semantic(Shader::USAGE_COLOR, 1);
+ if(dst.y) output[D1][1] = Semantic(Shader::USAGE_COLOR, 1);
+ if(dst.z) output[D1][2] = Semantic(Shader::USAGE_COLOR, 1);
+ if(dst.w) output[D1][3] = Semantic(Shader::USAGE_COLOR, 1);
}
else ASSERT(false);
break;
- case ShaderParameter::PARAMETER_TEXCRDOUT:
- if(dst.x) output[T0 + dst.index][0] = Semantic(ShaderOperation::USAGE_TEXCOORD, dst.index);
- if(dst.y) output[T0 + dst.index][1] = Semantic(ShaderOperation::USAGE_TEXCOORD, dst.index);
- if(dst.z) output[T0 + dst.index][2] = Semantic(ShaderOperation::USAGE_TEXCOORD, dst.index);
- if(dst.w) output[T0 + dst.index][3] = Semantic(ShaderOperation::USAGE_TEXCOORD, dst.index);
+ case Shader::PARAMETER_TEXCRDOUT:
+ if(dst.x) output[T0 + dst.index][0] = Semantic(Shader::USAGE_TEXCOORD, dst.index);
+ if(dst.y) output[T0 + dst.index][1] = Semantic(Shader::USAGE_TEXCOORD, dst.index);
+ if(dst.z) output[T0 + dst.index][2] = Semantic(Shader::USAGE_TEXCOORD, dst.index);
+ if(dst.w) output[T0 + dst.index][3] = Semantic(Shader::USAGE_TEXCOORD, dst.index);
break;
default:
break;
@@ -220,27 +227,27 @@
}
else // Shader Model 3.0 input declaration
{
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getOpcode() == ShaderOperation::OPCODE_DCL &&
- instruction[i]->getDestinationParameter().type == ShaderParameter::PARAMETER_OUTPUT)
+ if(instruction[i]->opcode == Shader::OPCODE_DCL &&
+ instruction[i]->dst.type == Shader::PARAMETER_OUTPUT)
{
- unsigned char usage = instruction[i]->getUsage();
- unsigned char usageIndex = instruction[i]->getUsageIndex();
+ unsigned char usage = instruction[i]->usage;
+ unsigned char usageIndex = instruction[i]->usageIndex;
- const Instruction::DestinationParameter &dst = instruction[i]->getDestinationParameter();
+ const DestinationParameter &dst = instruction[i]->dst;
if(dst.x) output[dst.index][0] = Semantic(usage, usageIndex);
if(dst.y) output[dst.index][1] = Semantic(usage, usageIndex);
if(dst.z) output[dst.index][2] = Semantic(usage, usageIndex);
if(dst.w) output[dst.index][3] = Semantic(usage, usageIndex);
- if(usage == ShaderOperation::USAGE_POSITION && usageIndex == 0)
+ if(usage == Shader::USAGE_POSITION && usageIndex == 0)
{
positionRegister = dst.index;
}
- if(usage == ShaderOperation::USAGE_PSIZE && usageIndex == 0)
+ if(usage == Shader::USAGE_PSIZE && usageIndex == 0)
{
pointSizeRegister = dst.index;
}
@@ -253,9 +260,9 @@
{
texldl = false;
- for(int i = 0; i < length; i++)
+ for(unsigned int i = 0; i < instruction.size(); i++)
{
- if(instruction[i]->getOpcode() == Instruction::Operation::OPCODE_TEXLDL)
+ if(instruction[i]->opcode == Shader::OPCODE_TEXLDL)
{
texldl = true;
diff --git a/src/Shader/VertexShader.hpp b/src/Shader/VertexShader.hpp
index 1200b20..3bfc0f4 100644
--- a/src/Shader/VertexShader.hpp
+++ b/src/Shader/VertexShader.hpp
@@ -1,48 +1,47 @@
-// SwiftShader Software Renderer
-//
-// Copyright(c) 2005-2011 TransGaming Inc.
-//
-// All rights reserved. No part of this software may be copied, distributed, transmitted,
-// transcribed, stored in a retrieval system, translated into any human or computer
-// language by any means, or disclosed to third parties without the explicit written
-// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
-// or implied, including but not limited to any patent rights, are granted to you.
-//
-
-#ifndef sw_VertexShader_hpp
-#define sw_VertexShader_hpp
-
-#include "Shader.hpp"
-
-namespace sw
-{
- class VertexShader : public Shader
- {
- public:
- VertexShader(const unsigned long *token);
-
- virtual ~VertexShader();
-
- static int validate(const unsigned long *const token); // Returns number of instructions if valid
- bool containsTexldl() const;
-
- int positionRegister; // FIXME: Private
- int pointSizeRegister; // FIXME: Private
-
- Semantic input[16]; // FIXME: Private
- Semantic output[12][4]; // FIXME: Private
-
- private:
- void parse(const unsigned long *token);
-
- void analyzeInput();
- void analyzeOutput();
- void analyzeTexldl();
-
- bool texldl;
- };
-
- typedef VertexShader::Instruction VertexShaderInstruction;
-}
-
-#endif // sw_VertexShader_hpp
+// SwiftShader Software Renderer
+//
+// Copyright(c) 2005-2012 TransGaming Inc.
+//
+// All rights reserved. No part of this software may be copied, distributed, transmitted,
+// transcribed, stored in a retrieval system, translated into any human or computer
+// language by any means, or disclosed to third parties without the explicit written
+// agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express
+// or implied, including but not limited to any patent rights, are granted to you.
+//
+
+#ifndef sw_VertexShader_hpp
+#define sw_VertexShader_hpp
+
+#include "Shader.hpp"
+
+namespace sw
+{
+ class VertexShader : public Shader
+ {
+ public:
+ explicit VertexShader(const VertexShader *vs = 0);
+ explicit VertexShader(const unsigned long *token);
+
+ virtual ~VertexShader();
+
+ static int validate(const unsigned long *const token); // Returns number of instructions if valid
+ bool containsTexldl() const;
+
+ virtual void analyze();
+
+ int positionRegister; // FIXME: Private
+ int pointSizeRegister; // FIXME: Private
+
+ Semantic input[16]; // FIXME: Private
+ Semantic output[12][4]; // FIXME: Private
+
+ private:
+ void analyzeInput();
+ void analyzeOutput();
+ void analyzeTexldl();
+
+ bool texldl;
+ };
+}
+
+#endif // sw_VertexShader_hpp
