ShuriZma
/
dolphin
mirror of https://github.com/dolphin-emu/dolphin.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #3055 from degasus/videosw 

Video Software cleanup
This commit is contained in:
Markus Wick 2015-09-17 18:42:17 +02:00
parent 52de4cc476 3b5b80bf3c
commit edae1d3bcb
13 changed files with 202 additions and 884 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
Source/Core/DolphinWX/SoftwareVideoConfigDialog.cpp
View File

@ -76,9 +76,6 @@ SoftwareVideoConfigDialog::SoftwareVideoConfigDialog(wxWindow* parent, const std
choice_backend->Disable(); choice_backend->Disable();
} }
// rasterizer
szr_rendering->Add(new SettingCheckBox(page_general, _("Hardware rasterization"), "", vconfig.bHwRasterizer));
// xfb // xfb
szr_rendering->Add(new SettingCheckBox(page_general, _("Bypass XFB"), "", vconfig.bBypassXFB)); szr_rendering->Add(new SettingCheckBox(page_general, _("Bypass XFB"), "", vconfig.bBypassXFB));
} }

1
Source/Core/VideoBackends/Software/CMakeLists.txt
View File

@ -5,7 +5,6 @@ set(SRCS BPMemLoader.cpp
DebugUtil.cpp DebugUtil.cpp
EfbCopy.cpp EfbCopy.cpp
EfbInterface.cpp EfbInterface.cpp
HwRasterizer.cpp
SWmain.cpp SWmain.cpp
OpcodeDecoder.cpp OpcodeDecoder.cpp
RasterFont.cpp RasterFont.cpp

15
Source/Core/VideoBackends/Software/DebugUtil.cpp
View File

@ -11,7 +11,6 @@
#include "VideoBackends/Software/BPMemLoader.h" #include "VideoBackends/Software/BPMemLoader.h"
#include "VideoBackends/Software/DebugUtil.h" #include "VideoBackends/Software/DebugUtil.h"
#include "VideoBackends/Software/EfbInterface.h" #include "VideoBackends/Software/EfbInterface.h"
#include "VideoBackends/Software/HwRasterizer.h"
#include "VideoBackends/Software/SWCommandProcessor.h" #include "VideoBackends/Software/SWCommandProcessor.h"
#include "VideoBackends/Software/SWRenderer.h" #include "VideoBackends/Software/SWRenderer.h"
#include "VideoBackends/Software/SWStatistics.h" #include "VideoBackends/Software/SWStatistics.h"
@ -25,8 +24,6 @@
namespace DebugUtil namespace DebugUtil
{ {
static bool drawingHwTriangles = false;
static const int NUM_OBJECT_BUFFERS = 40; static const int NUM_OBJECT_BUFFERS = 40;
static u32 *ObjectBuffer[NUM_OBJECT_BUFFERS]; static u32 *ObjectBuffer[NUM_OBJECT_BUFFERS];
@ -207,12 +204,6 @@ void OnObjectBegin()
{ {
if (g_SWVideoConfig.bDumpTextures && swstats.thisFrame.numDrawnObjects >= g_SWVideoConfig.drawStart && swstats.thisFrame.numDrawnObjects < g_SWVideoConfig.drawEnd) if (g_SWVideoConfig.bDumpTextures && swstats.thisFrame.numDrawnObjects >= g_SWVideoConfig.drawStart && swstats.thisFrame.numDrawnObjects < g_SWVideoConfig.drawEnd)
DumpActiveTextures(); DumpActiveTextures();
if (g_SWVideoConfig.bHwRasterizer)
{
HwRasterizer::BeginTriangles();
drawingHwTriangles = true;
}
} }
} }
@ -225,12 +216,6 @@ void OnObjectEnd()
File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), File::GetUserPath(D_DUMPFRAMES_IDX).c_str(),
swstats.thisFrame.numDrawnObjects)); swstats.thisFrame.numDrawnObjects));
if (g_SWVideoConfig.bHwRasterizer || drawingHwTriangles)
{
HwRasterizer::EndTriangles();
drawingHwTriangles = false;
}
for (int i = 0; i < NUM_OBJECT_BUFFERS; i++) for (int i = 0; i < NUM_OBJECT_BUFFERS; i++)
{ {
if (DrawnToBuffer[i]) if (DrawnToBuffer[i])

60
Source/Core/VideoBackends/Software/EfbCopy.cpp
View File

@ -9,7 +9,6 @@
#include "VideoBackends/Software/DebugUtil.h" #include "VideoBackends/Software/DebugUtil.h"
#include "VideoBackends/Software/EfbCopy.h" #include "VideoBackends/Software/EfbCopy.h"
#include "VideoBackends/Software/EfbInterface.h" #include "VideoBackends/Software/EfbInterface.h"
#include "VideoBackends/Software/HwRasterizer.h"
#include "VideoBackends/Software/SWCommandProcessor.h" #include "VideoBackends/Software/SWCommandProcessor.h"
#include "VideoBackends/Software/SWRenderer.h" #include "VideoBackends/Software/SWRenderer.h"
#include "VideoBackends/Software/SWStatistics.h" #include "VideoBackends/Software/SWStatistics.h"
@ -31,47 +30,41 @@ namespace EfbCopy
{ {
GLInterface->Update(); // update the render window position and the backbuffer size GLInterface->Update(); // update the render window position and the backbuffer size
if (!g_SWVideoConfig.bHwRasterizer) INFO_LOG(VIDEO, "xfbaddr: %x, fbwidth: %i, fbheight: %i, source: (%i, %i, %i, %i), Gamma %f",
xfbAddr, fbWidth, fbHeight, sourceRc.top, sourceRc.left, sourceRc.bottom, sourceRc.right, Gamma);
if (!g_SWVideoConfig.bBypassXFB)
{ {
INFO_LOG(VIDEO, "xfbaddr: %x, fbwidth: %i, fbheight: %i, source: (%i, %i, %i, %i), Gamma %f", EfbInterface::yuv422_packed* xfb_in_ram = (EfbInterface::yuv422_packed *) Memory::GetPointer(xfbAddr);
xfbAddr, fbWidth, fbHeight, sourceRc.top, sourceRc.left, sourceRc.bottom, sourceRc.right, Gamma);
if (!g_SWVideoConfig.bBypassXFB) EfbInterface::CopyToXFB(xfb_in_ram, fbWidth, fbHeight, sourceRc, Gamma);
{ }
EfbInterface::yuv422_packed* xfb_in_ram = (EfbInterface::yuv422_packed *) Memory::GetPointer(xfbAddr); else
{
// Ask SWRenderer for the next color texture
u8 *colorTexture = SWRenderer::GetNextColorTexture();
EfbInterface::CopyToXFB(xfb_in_ram, fbWidth, fbHeight, sourceRc, Gamma); EfbInterface::BypassXFB(colorTexture, fbWidth, fbHeight, sourceRc, Gamma);
}
else
{
// Ask SWRenderer for the next color texture
u8 *colorTexture = SWRenderer::GetNextColorTexture();
EfbInterface::BypassXFB(colorTexture, fbWidth, fbHeight, sourceRc, Gamma); // Tell SWRenderer we are now finished with it.
SWRenderer::SwapColorTexture();
// Tell SWRenderer we are now finished with it. // FifoPlayer is broken and never calls BeginFrame/EndFrame.
SWRenderer::SwapColorTexture(); // Hence, we manually force a swap now. This emulates the behavior
// of hardware backends with XFB emulation disabled.
// FifoPlayer is broken and never calls BeginFrame/EndFrame. // TODO: Fix FifoPlayer by making proper use of VideoInterface!
// Hence, we manually force a swap now. This emulates the behavior // This requires careful synchronization since GPU commands
// of hardware backends with XFB emulation disabled. // are processed on a different thread than VI commands.
// TODO: Fix FifoPlayer by making proper use of VideoInterface! SWRenderer::Swap(fbWidth, fbHeight);
// This requires careful synchronization since GPU commands DebugUtil::OnFrameEnd(fbWidth, fbHeight);
// are processed on a different thread than VI commands.
SWRenderer::Swap(fbWidth, fbHeight);
DebugUtil::OnFrameEnd(fbWidth, fbHeight);
}
} }
} }
static void CopyToRam() static void CopyToRam()
{ {
if (!g_SWVideoConfig.bHwRasterizer) u8 *dest_ptr = Memory::GetPointer(bpmem.copyTexDest << 5);
{
u8 *dest_ptr = Memory::GetPointer(bpmem.copyTexDest << 5);
TextureEncoder::Encode(dest_ptr); TextureEncoder::Encode(dest_ptr);
}
} }
static void ClearEfb() static void ClearEfb()
@ -138,10 +131,7 @@ namespace EfbCopy
if (bpmem.triggerEFBCopy.clear) if (bpmem.triggerEFBCopy.clear)
{ {
if (g_SWVideoConfig.bHwRasterizer) ClearEfb();
HwRasterizer::Clear();
else
ClearEfb();
} }
} }
} }

382
Source/Core/VideoBackends/Software/HwRasterizer.cpp
View File

@ -1,382 +0,0 @@
// Copyright 2009 Dolphin Emulator Project
// Licensed under GPLv2+
// Refer to the license.txt file included.
#include "Common/CommonTypes.h"
#include "Common/MemoryUtil.h"
#include "VideoBackends/OGL/GLInterfaceBase.h"
#include "VideoBackends/Software/BPMemLoader.h"
#include "VideoBackends/Software/DebugUtil.h"
#include "VideoBackends/Software/HwRasterizer.h"
#include "VideoBackends/Software/NativeVertexFormat.h"
#include "VideoCommon/VideoCommon.h"
#define TEMP_SIZE (1024*1024*4)
namespace HwRasterizer
{
static float efbHalfWidth;
static float efbHalfHeight;
static bool hasTexture;
static u8 *temp;
// Programs
static GLuint colProg, texProg, clearProg;
// Texture type
static GLenum texType;
// Color
static GLint col_apos = -1, col_atex = -1;
// Tex
static GLint tex_apos = -1, tex_atex = -1, tex_utex = -1;
// Clear shader
static GLint clear_apos = -1, clear_ucol = -1;
static void CreateShaders()
{
// Color Vertices
static const char *fragcolText =
"#ifdef GL_ES\n"
"precision highp float;\n"
"#endif\n"
"varying vec4 TexCoordOut;\n"
"void main() {\n"
" gl_FragColor = TexCoordOut;\n"
"}\n";
// Texture Vertices
static const char *fragtexText =
"#ifdef GL_ES\n"
"precision highp float;\n"
"#define texture2DRect texture2D\n"
"#define sampler2DRect sampler2D\n"
"#endif\n"
"varying vec4 TexCoordOut;\n"
"uniform sampler2DRect Texture;\n"
"void main() {\n"
" gl_FragColor = texture2DRect(Texture, TexCoordOut.xy);\n"
"}\n";
// Clear shader
static const char *fragclearText =
"#ifdef GL_ES\n"
"precision highp float;\n"
"#endif\n"
"uniform vec4 Color;\n"
"void main() {\n"
" gl_FragColor = Color;\n"
"}\n";
// Generic passthrough vertice shaders
static const char *vertShaderText =
"attribute vec4 pos;\n"
"attribute vec4 TexCoordIn;\n "
"varying vec4 TexCoordOut;\n "
"void main() {\n"
" gl_Position = pos;\n"
" TexCoordOut = TexCoordIn;\n"
"}\n";
static const char *vertclearText =
"attribute vec4 pos;\n"
"void main() {\n"
" gl_Position = pos;\n"
"}\n";
// Color Program
colProg = OpenGL_CompileProgram(vertShaderText, fragcolText);
// Texture Program
texProg = OpenGL_CompileProgram(vertShaderText, fragtexText);
// Clear Program
clearProg = OpenGL_CompileProgram(vertclearText, fragclearText);
// Color attributes
col_apos = glGetAttribLocation(colProg, "pos");
col_atex = glGetAttribLocation(colProg, "TexCoordIn");
// Texture attributes
tex_apos = glGetAttribLocation(texProg, "pos");
tex_atex = glGetAttribLocation(texProg, "TexCoordIn");
tex_utex = glGetUniformLocation(texProg, "Texture");
// Clear attributes
clear_apos = glGetAttribLocation(clearProg, "pos");
clear_ucol = glGetUniformLocation(clearProg, "Color");
}
void Init()
{
efbHalfWidth = EFB_WIDTH / 2.0f;
efbHalfHeight = 480 / 2.0f;
temp = (u8*)AllocateMemoryPages(TEMP_SIZE);
}
void Shutdown()
{
glDeleteProgram(colProg);
glDeleteProgram(texProg);
glDeleteProgram(clearProg);
}
void Prepare()
{
//legacy multitexturing: select texture channel only.
glActiveTexture(GL_TEXTURE0);
glPixelStorei(GL_UNPACK_ALIGNMENT, 4); // 4-byte pixel alignment
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGL)
{
glShadeModel(GL_SMOOTH);
glDisable(GL_BLEND);
glClearDepth(1.0f);
glEnable(GL_SCISSOR_TEST);
glDisable(GL_LIGHTING);
glMatrixMode(GL_PROJECTION);
glLoadIdentity();
glMatrixMode(GL_MODELVIEW);
glLoadIdentity();
glClientActiveTexture(GL_TEXTURE0);
glTexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
glEnable(GL_TEXTURE_RECTANGLE_ARB);
glStencilFunc(GL_ALWAYS, 0, 0);
glDisable(GL_STENCIL_TEST);
}
// used by hw rasterizer if it enables blending and depth test
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
glDepthFunc(GL_LEQUAL);
glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
CreateShaders();
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGL)
texType = GL_TEXTURE_RECTANGLE;
else
texType = GL_TEXTURE_2D;
}
static float width, height;
static void LoadTexture()
{
FourTexUnits &texUnit = bpmem.tex[0];
u32 imageAddr = texUnit.texImage3[0].image_base;
// Texture Rectangle uses pixel coordinates
// While GLES uses texture coordinates
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGL)
{
width = (float)texUnit.texImage0[0].width;
height = (float)texUnit.texImage0[0].height;
}
else
{
width = 1;
height = 1;
}
TexCacheEntry &cacheEntry = textures[imageAddr];
cacheEntry.Update();
glBindTexture(texType, cacheEntry.texture);
glTexParameteri(texType, GL_TEXTURE_MAG_FILTER, texUnit.texMode0[0].mag_filter ? GL_LINEAR : GL_NEAREST);
glTexParameteri(texType, GL_TEXTURE_MIN_FILTER, (texUnit.texMode0[0].min_filter >= 4) ? GL_LINEAR : GL_NEAREST);
}
void BeginTriangles()
{
// disabling depth test sometimes allows more things to be visible
glEnable(GL_DEPTH_TEST);
glEnable(GL_BLEND);
hasTexture = bpmem.tevorders[0].enable0;
if (hasTexture)
LoadTexture();
}
void EndTriangles()
{
glBindTexture(texType, 0);
glDisable(GL_DEPTH_TEST);
glDisable(GL_BLEND);
}
static void DrawColorVertex(OutputVertexData *v0, OutputVertexData *v1, OutputVertexData *v2)
{
float x0 = (v0->screenPosition.x / efbHalfWidth) - 1.0f;
float y0 = 1.0f - (v0->screenPosition.y / efbHalfHeight);
float z0 = v0->screenPosition.z / (float)0x00ffffff;
float x1 = (v1->screenPosition.x / efbHalfWidth) - 1.0f;
float y1 = 1.0f - (v1->screenPosition.y / efbHalfHeight);
float z1 = v1->screenPosition.z / (float)0x00ffffff;
float x2 = (v2->screenPosition.x / efbHalfWidth) - 1.0f;
float y2 = 1.0f - (v2->screenPosition.y / efbHalfHeight);
float z2 = v2->screenPosition.z / (float)0x00ffffff;
float r0 = v0->color[0][OutputVertexData::RED_C] / 255.0f;
float g0 = v0->color[0][OutputVertexData::GRN_C] / 255.0f;
float b0 = v0->color[0][OutputVertexData::BLU_C] / 255.0f;
float r1 = v1->color[0][OutputVertexData::RED_C] / 255.0f;
float g1 = v1->color[0][OutputVertexData::GRN_C] / 255.0f;
float b1 = v1->color[0][OutputVertexData::BLU_C] / 255.0f;
float r2 = v2->color[0][OutputVertexData::RED_C] / 255.0f;
float g2 = v2->color[0][OutputVertexData::GRN_C] / 255.0f;
float b2 = v2->color[0][OutputVertexData::BLU_C] / 255.0f;
static const GLfloat verts[3][3] = {
{ x0, y0, z0 },
{ x1, y1, z1 },
{ x2, y2, z2 }
};
static const GLfloat col[3][4] = {
{ r0, g0, b0, 1.0f },
{ r1, g1, b1, 1.0f },
{ r2, g2, b2, 1.0f }
};
{
glUseProgram(colProg);
glEnableVertexAttribArray(col_apos);
glEnableVertexAttribArray(col_atex);
glVertexAttribPointer(col_apos, 3, GL_FLOAT, GL_FALSE, 0, verts);
glVertexAttribPointer(col_atex, 4, GL_FLOAT, GL_FALSE, 0, col);
glDrawArrays(GL_TRIANGLES, 0, 3);
glDisableVertexAttribArray(col_atex);
glDisableVertexAttribArray(col_apos);
}
}
static void DrawTextureVertex(OutputVertexData *v0, OutputVertexData *v1, OutputVertexData *v2)
{
float x0 = (v0->screenPosition.x / efbHalfWidth) - 1.0f;
float y0 = 1.0f - (v0->screenPosition.y / efbHalfHeight);
float z0 = v0->screenPosition.z;
float x1 = (v1->screenPosition.x / efbHalfWidth) - 1.0f;
float y1 = 1.0f - (v1->screenPosition.y / efbHalfHeight);
float z1 = v1->screenPosition.z;
float x2 = (v2->screenPosition.x / efbHalfWidth) - 1.0f;
float y2 = 1.0f - (v2->screenPosition.y / efbHalfHeight);
float z2 = v2->screenPosition.z;
float s0 = v0->texCoords[0].x / width;
float t0 = v0->texCoords[0].y / height;
float s1 = v1->texCoords[0].x / width;
float t1 = v1->texCoords[0].y / height;
float s2 = v2->texCoords[0].x / width;
float t2 = v2->texCoords[0].y / height;
static const GLfloat verts[3][3] = {
{ x0, y0, z0 },
{ x1, y1, z1 },
{ x2, y2, z2 }
};
static const GLfloat tex[3][2] = {
{ s0, t0 },
{ s1, t1 },
{ s2, t2 }
};
{
glUseProgram(texProg);
glEnableVertexAttribArray(tex_apos);
glEnableVertexAttribArray(tex_atex);
glVertexAttribPointer(tex_apos, 3, GL_FLOAT, GL_FALSE, 0, verts);
glVertexAttribPointer(tex_atex, 2, GL_FLOAT, GL_FALSE, 0, tex);
glUniform1i(tex_utex, 0);
glDrawArrays(GL_TRIANGLES, 0, 3);
glDisableVertexAttribArray(tex_atex);
glDisableVertexAttribArray(tex_apos);
}
}
void DrawTriangleFrontFace(OutputVertexData *v0, OutputVertexData *v1, OutputVertexData *v2)
{
if (hasTexture)
DrawTextureVertex(v0, v1, v2);
else
DrawColorVertex(v0, v1, v2);
}
void Clear()
{
u8 r = (bpmem.clearcolorAR & 0x00ff);
u8 g = (bpmem.clearcolorGB & 0xff00) >> 8;
u8 b = (bpmem.clearcolorGB & 0x00ff);
u8 a = (bpmem.clearcolorAR & 0xff00) >> 8;
GLfloat left = (GLfloat)bpmem.copyTexSrcXY.x / efbHalfWidth - 1.0f;
GLfloat top = 1.0f - (GLfloat)bpmem.copyTexSrcXY.y / efbHalfHeight;
GLfloat right = (GLfloat)(left + bpmem.copyTexSrcWH.x + 1) / efbHalfWidth - 1.0f;
GLfloat bottom = 1.0f - (GLfloat)(top + bpmem.copyTexSrcWH.y + 1) / efbHalfHeight;
GLfloat depth = (GLfloat)bpmem.clearZValue / (GLfloat)0x00ffffff;
static const GLfloat verts[4][3] = {
{ left, top, depth },
{ right, top, depth },
{ right, bottom, depth },
{ left, bottom, depth }
};
{
glUseProgram(clearProg);
glVertexAttribPointer(clear_apos, 3, GL_FLOAT, GL_FALSE, 0, verts);
glUniform4f(clear_ucol, r, g, b, a);
glEnableVertexAttribArray(col_apos);
glDrawArrays(GL_TRIANGLE_FAN, 0, 4);
glDisableVertexAttribArray(col_apos);
}
}
TexCacheEntry::TexCacheEntry()
{
Create();
}
void TexCacheEntry::Create()
{
FourTexUnits &texUnit = bpmem.tex[0];
texImage0.hex = texUnit.texImage0[0].hex;
texImage1.hex = texUnit.texImage1[0].hex;
texImage2.hex = texUnit.texImage2[0].hex;
texImage3.hex = texUnit.texImage3[0].hex;
texTlut.hex = texUnit.texTlut[0].hex;
int image_width = texImage0.width;
int image_height = texImage0.height;
DebugUtil::GetTextureRGBA(temp, 0, 0, image_width, image_height);
glGenTextures(1, (GLuint *)&texture);
glBindTexture(texType, texture);
glTexImage2D(texType, 0, GL_RGBA, (GLsizei)image_width, (GLsizei)image_height, 0, GL_RGBA, GL_UNSIGNED_BYTE, temp);
}
void TexCacheEntry::Destroy()
{
if (texture == 0)
return;
glDeleteTextures(1, &texture);
texture = 0;
}
void TexCacheEntry::Update()
{
FourTexUnits &texUnit = bpmem.tex[0];
// extra checks cause textures to be reloaded much more
if (texUnit.texImage0[0].hex != texImage0.hex ||
// texUnit.texImage1[0].hex != texImage1.hex ||
// texUnit.texImage2[0].hex != texImage2.hex ||
texUnit.texImage3[0].hex != texImage3.hex ||
texUnit.texTlut[0].hex != texTlut.hex)
{
Destroy();
Create();
}
}
}

47
Source/Core/VideoBackends/Software/HwRasterizer.h
View File

@ -1,47 +0,0 @@
// Copyright 2009 Dolphin Emulator Project
// Licensed under GPLv2+
// Refer to the license.txt file included.
#pragma once
#include <map>
#include "VideoBackends/OGL/GLUtil.h"
#include "VideoBackends/Software/BPMemLoader.h"
struct OutputVertexData;
namespace HwRasterizer
{
void Init();
void Shutdown();
void Prepare();
void BeginTriangles();
void EndTriangles();
void DrawTriangleFrontFace(OutputVertexData *v0, OutputVertexData *v1, OutputVertexData *v2);
void Clear();
struct TexCacheEntry
{
TexImage0 texImage0;
TexImage1 texImage1;
TexImage2 texImage2;
TexImage3 texImage3;
TexTLUT texTlut;
GLuint texture;
TexCacheEntry();
void Create();
void Destroy();
void Update();
};
typedef std::map<u32, TexCacheEntry> TextureCache;
static TextureCache textures;
}

338
Source/Core/VideoBackends/Software/Rasterizer.cpp
View File

@ -7,7 +7,6 @@
#include "Common/CommonTypes.h" #include "Common/CommonTypes.h"
#include "VideoBackends/Software/BPMemLoader.h" #include "VideoBackends/Software/BPMemLoader.h"
#include "VideoBackends/Software/EfbInterface.h" #include "VideoBackends/Software/EfbInterface.h"
#include "VideoBackends/Software/HwRasterizer.h"
#include "VideoBackends/Software/NativeVertexFormat.h" #include "VideoBackends/Software/NativeVertexFormat.h"
#include "VideoBackends/Software/Rasterizer.h" #include "VideoBackends/Software/Rasterizer.h"
#include "VideoBackends/Software/SWStatistics.h" #include "VideoBackends/Software/SWStatistics.h"
@ -129,7 +128,7 @@ inline void Draw(s32 x, s32 y, s32 xi, s32 yi)
s32 z = (s32)MathUtil::Clamp<float>(ZSlope.GetValue(dx, dy), 0.0f, 16777215.0f); s32 z = (s32)MathUtil::Clamp<float>(ZSlope.GetValue(dx, dy), 0.0f, 16777215.0f);
if (!BoundingBox::active && bpmem.UseEarlyDepthTest() && g_SWVideoConfig.bZComploc) if (bpmem.UseEarlyDepthTest() && g_SWVideoConfig.bZComploc)
{ {
// TODO: Test if perf regs are incremented even if test is disabled // TODO: Test if perf regs are incremented even if test is disabled
EfbInterface::IncPerfCounterQuadCount(PQ_ZCOMP_INPUT_ZCOMPLOC); EfbInterface::IncPerfCounterQuadCount(PQ_ZCOMP_INPUT_ZCOMPLOC);
@ -332,12 +331,6 @@ void DrawTriangleFrontFace(OutputVertexData *v0, OutputVertexData *v1, OutputVer
{ {
INCSTAT(swstats.thisFrame.numTrianglesDrawn); INCSTAT(swstats.thisFrame.numTrianglesDrawn);
if (g_SWVideoConfig.bHwRasterizer && !BoundingBox::active)
{
HwRasterizer::DrawTriangleFrontFace(v0, v1, v2);
return;
}
// adapted from http://devmaster.net/posts/6145/advanced-rasterization // adapted from http://devmaster.net/posts/6145/advanced-rasterization
// 28.4 fixed-pou32 coordinates. rounded to nearest and adjusted to match hardware output // 28.4 fixed-pou32 coordinates. rounded to nearest and adjusted to match hardware output
@ -425,281 +418,88 @@ void DrawTriangleFrontFace(OutputVertexData *v0, OutputVertexData *v1, OutputVer
if (DY23 < 0 || (DY23 == 0 && DX23 > 0)) C2++; if (DY23 < 0 || (DY23 == 0 && DX23 > 0)) C2++;
if (DY31 < 0 || (DY31 == 0 && DX31 > 0)) C3++; if (DY31 < 0 || (DY31 == 0 && DX31 > 0)) C3++;
// If drawing, rasterize every block // Start in corner of 8x8 block
if (!BoundingBox::active) minx &= ~(BLOCK_SIZE - 1);
miny &= ~(BLOCK_SIZE - 1);
// Loop through blocks
for (s32 y = miny; y < maxy; y += BLOCK_SIZE)
{ {
// Start in corner of 8x8 block for (s32 x = minx; x < maxx; x += BLOCK_SIZE)
minx &= ~(BLOCK_SIZE - 1);
miny &= ~(BLOCK_SIZE - 1);
// Loop through blocks
for (s32 y = miny; y < maxy; y += BLOCK_SIZE)
{ {
for (s32 x = minx; x < maxx; x += BLOCK_SIZE) // Corners of block
s32 x0 = x << 4;
s32 x1 = (x + BLOCK_SIZE - 1) << 4;
s32 y0 = y << 4;
s32 y1 = (y + BLOCK_SIZE - 1) << 4;
// Evaluate half-space functions
bool a00 = C1 + DX12 * y0 - DY12 * x0 > 0;
bool a10 = C1 + DX12 * y0 - DY12 * x1 > 0;
bool a01 = C1 + DX12 * y1 - DY12 * x0 > 0;
bool a11 = C1 + DX12 * y1 - DY12 * x1 > 0;
int a = (a00 << 0) | (a10 << 1) | (a01 << 2) | (a11 << 3);
bool b00 = C2 + DX23 * y0 - DY23 * x0 > 0;
bool b10 = C2 + DX23 * y0 - DY23 * x1 > 0;
bool b01 = C2 + DX23 * y1 - DY23 * x0 > 0;
bool b11 = C2 + DX23 * y1 - DY23 * x1 > 0;
int b = (b00 << 0) | (b10 << 1) | (b01 << 2) | (b11 << 3);
bool c00 = C3 + DX31 * y0 - DY31 * x0 > 0;
bool c10 = C3 + DX31 * y0 - DY31 * x1 > 0;
bool c01 = C3 + DX31 * y1 - DY31 * x0 > 0;
bool c11 = C3 + DX31 * y1 - DY31 * x1 > 0;
int c = (c00 << 0) | (c10 << 1) | (c01 << 2) | (c11 << 3);
// Skip block when outside an edge
if (a == 0x0 || b == 0x0 || c == 0x0)
continue;
BuildBlock(x, y);
// Accept whole block when totally covered
if (a == 0xF && b == 0xF && c == 0xF)
{ {
// Corners of block for (s32 iy = 0; iy < BLOCK_SIZE; iy++)
s32 x0 = x << 4;
s32 x1 = (x + BLOCK_SIZE - 1) << 4;
s32 y0 = y << 4;
s32 y1 = (y + BLOCK_SIZE - 1) << 4;
// Evaluate half-space functions
bool a00 = C1 + DX12 * y0 - DY12 * x0 > 0;
bool a10 = C1 + DX12 * y0 - DY12 * x1 > 0;
bool a01 = C1 + DX12 * y1 - DY12 * x0 > 0;
bool a11 = C1 + DX12 * y1 - DY12 * x1 > 0;
int a = (a00 << 0) | (a10 << 1) | (a01 << 2) | (a11 << 3);
bool b00 = C2 + DX23 * y0 - DY23 * x0 > 0;
bool b10 = C2 + DX23 * y0 - DY23 * x1 > 0;
bool b01 = C2 + DX23 * y1 - DY23 * x0 > 0;
bool b11 = C2 + DX23 * y1 - DY23 * x1 > 0;
int b = (b00 << 0) | (b10 << 1) | (b01 << 2) | (b11 << 3);
bool c00 = C3 + DX31 * y0 - DY31 * x0 > 0;
bool c10 = C3 + DX31 * y0 - DY31 * x1 > 0;
bool c01 = C3 + DX31 * y1 - DY31 * x0 > 0;
bool c11 = C3 + DX31 * y1 - DY31 * x1 > 0;
int c = (c00 << 0) | (c10 << 1) | (c01 << 2) | (c11 << 3);
// Skip block when outside an edge
if (a == 0x0 || b == 0x0 || c == 0x0)
continue;
BuildBlock(x, y);
// Accept whole block when totally covered
if (a == 0xF && b == 0xF && c == 0xF)
{ {
for (s32 iy = 0; iy < BLOCK_SIZE; iy++) for (s32 ix = 0; ix < BLOCK_SIZE; ix++)
{ {
for (s32 ix = 0; ix < BLOCK_SIZE; ix++) Draw(x + ix, y + iy, ix, iy);
}
}
}
else // Partially covered block
{
s32 CY1 = C1 + DX12 * y0 - DY12 * x0;
s32 CY2 = C2 + DX23 * y0 - DY23 * x0;
s32 CY3 = C3 + DX31 * y0 - DY31 * x0;
for (s32 iy = 0; iy < BLOCK_SIZE; iy++)
{
s32 CX1 = CY1;
s32 CX2 = CY2;
s32 CX3 = CY3;
for (s32 ix = 0; ix < BLOCK_SIZE; ix++)
{
if (CX1 > 0 && CX2 > 0 && CX3 > 0)
{ {
Draw(x + ix, y + iy, ix, iy); Draw(x + ix, y + iy, ix, iy);
} }
CX1 -= FDY12;
CX2 -= FDY23;
CX3 -= FDY31;
} }
}
else // Partially covered block
{
s32 CY1 = C1 + DX12 * y0 - DY12 * x0;
s32 CY2 = C2 + DX23 * y0 - DY23 * x0;
s32 CY3 = C3 + DX31 * y0 - DY31 * x0;
for (s32 iy = 0; iy < BLOCK_SIZE; iy++) CY1 += FDX12;
{ CY2 += FDX23;
s32 CX1 = CY1; CY3 += FDX31;
s32 CX2 = CY2;
s32 CX3 = CY3;
for (s32 ix = 0; ix < BLOCK_SIZE; ix++)
{
if (CX1 > 0 && CX2 > 0 && CX3 > 0)
{
Draw(x + ix, y + iy, ix, iy);
}
CX1 -= FDY12;
CX2 -= FDY23;
CX3 -= FDY31;
}
CY1 += FDX12;
CY2 += FDX23;
CY3 += FDX31;
}
} }
} }
} }
} }
else
{
// Calculating bbox
// First check for alpha channel - don't do anything it if always fails,
// Change bbox to primitive size if it always passes
AlphaTest::TEST_RESULT alphaRes = bpmem.alpha_test.TestResult();
if (alphaRes != AlphaTest::UNDETERMINED)
{
if (alphaRes == AlphaTest::PASS)
{
BoundingBox::coords[BoundingBox::TOP] = std::min(BoundingBox::coords[BoundingBox::TOP], (u16) miny);
BoundingBox::coords[BoundingBox::LEFT] = std::min(BoundingBox::coords[BoundingBox::LEFT], (u16) minx);
BoundingBox::coords[BoundingBox::BOTTOM] = std::max(BoundingBox::coords[BoundingBox::BOTTOM], (u16) maxy);
BoundingBox::coords[BoundingBox::RIGHT] = std::max(BoundingBox::coords[BoundingBox::RIGHT], (u16) maxx);
}
return;
}
// If we are calculating bbox with alpha, we only need to find the
// topmost, leftmost, bottom most and rightmost pixels to be drawn.
// So instead of drawing every single one of the triangle's pixels,
// four loops are run: one for the top pixel, one for the left, one for
// the bottom and one for the right. As soon as a pixel that is to be
// drawn is found, the loop breaks. This enables a ~150% speedbost in
// bbox calculation, albeit at the cost of some ugly repetitive code.
const s32 FLEFT = minx << 4;
const s32 FRIGHT = maxx << 4;
s32 FTOP = miny << 4;
s32 FBOTTOM = maxy << 4;
// Start checking for bbox top
s32 CY1 = C1 + DX12 * FTOP - DY12 * FLEFT;
s32 CY2 = C2 + DX23 * FTOP - DY23 * FLEFT;
s32 CY3 = C3 + DX31 * FTOP - DY31 * FLEFT;
// Loop
for (s32 y = miny; y <= maxy; ++y)
{
if (y >= BoundingBox::coords[BoundingBox::TOP])
break;
s32 CX1 = CY1;
s32 CX2 = CY2;
s32 CX3 = CY3;
for (s32 x = minx; x <= maxx; ++x)
{
if (CX1 > 0 && CX2 > 0 && CX3 > 0)
{
// Build the new raster block every other pixel
PrepareBlock(x, y);
Draw(x, y, x & (BLOCK_SIZE - 1), y & (BLOCK_SIZE - 1));
if (y >= BoundingBox::coords[BoundingBox::TOP])
break;
}
CX1 -= FDY12;
CX2 -= FDY23;
CX3 -= FDY31;
}
CY1 += FDX12;
CY2 += FDX23;
CY3 += FDX31;
}
// Update top limit
miny = std::max((s32) BoundingBox::coords[BoundingBox::TOP], miny);
FTOP = miny << 4;
// Checking for bbox left
s32 CX1 = C1 + DX12 * FTOP - DY12 * FLEFT;
s32 CX2 = C2 + DX23 * FTOP - DY23 * FLEFT;
s32 CX3 = C3 + DX31 * FTOP - DY31 * FLEFT;
// Loop
for (s32 x = minx; x <= maxx; ++x)
{
if (x >= BoundingBox::coords[BoundingBox::LEFT])
break;
CY1 = CX1;
CY2 = CX2;
CY3 = CX3;
for (s32 y = miny; y <= maxy; ++y)
{
if (CY1 > 0 && CY2 > 0 && CY3 > 0)
{
PrepareBlock(x, y);
Draw(x, y, x & (BLOCK_SIZE - 1), y & (BLOCK_SIZE - 1));
if (x >= BoundingBox::coords[BoundingBox::LEFT])
break;
}
CY1 += FDX12;
CY2 += FDX23;
CY3 += FDX31;
}
CX1 -= FDY12;
CX2 -= FDY23;
CX3 -= FDY31;
}
// Update left limit
minx = std::max((s32) BoundingBox::coords[BoundingBox::LEFT], minx);
// Checking for bbox bottom
CY1 = C1 + DX12 * FBOTTOM - DY12 * FRIGHT;
CY2 = C2 + DX23 * FBOTTOM - DY23 * FRIGHT;
CY3 = C3 + DX31 * FBOTTOM - DY31 * FRIGHT;
// Loop
for (s32 y = maxy; y >= miny; --y)
{
CX1 = CY1;
CX2 = CY2;
CX3 = CY3;
if (y <= BoundingBox::coords[BoundingBox::BOTTOM])
break;
for (s32 x = maxx; x >= minx; --x)
{
if (CX1 > 0 && CX2 > 0 && CX3 > 0)
{
// Build the new raster block every other pixel
PrepareBlock(x, y);
Draw(x, y, x & (BLOCK_SIZE - 1), y & (BLOCK_SIZE - 1));
if (y <= BoundingBox::coords[BoundingBox::BOTTOM])
break;
}
CX1 += FDY12;
CX2 += FDY23;
CX3 += FDY31;
}
CY1 -= FDX12;
CY2 -= FDX23;
CY3 -= FDX31;
}
// Update bottom limit
maxy = std::min((s32) BoundingBox::coords[BoundingBox::BOTTOM], maxy);
FBOTTOM = maxy << 4;
// Checking for bbox right
CX1 = C1 + DX12 * FBOTTOM - DY12 * FRIGHT;
CX2 = C2 + DX23 * FBOTTOM - DY23 * FRIGHT;
CX3 = C3 + DX31 * FBOTTOM - DY31 * FRIGHT;
// Loop
for (s32 x = maxx; x >= minx; --x)
{
if (x <= BoundingBox::coords[BoundingBox::RIGHT])
break;
CY1 = CX1;
CY2 = CX2;
CY3 = CX3;
for (s32 y = maxy; y >= miny; --y)
{
if (CY1 > 0 && CY2 > 0 && CY3 > 0)
{
// Build the new raster block every other pixel
PrepareBlock(x, y);
Draw(x, y, x & (BLOCK_SIZE - 1), y & (BLOCK_SIZE - 1));
if (x <= BoundingBox::coords[BoundingBox::RIGHT])
break;
}
CY1 -= FDX12;
CY2 -= FDX23;
CY3 -= FDX31;
}
CX1 += FDY12;
CX2 += FDY23;
CX3 += FDY31;
}
}
} }

3
Source/Core/VideoBackends/Software/SWRenderer.cpp
View File

@ -211,8 +211,7 @@ void SWRenderer::UpdateColorTexture(EfbInterface::yuv422_packed *xfb, u32 fbWidt
void SWRenderer::Swap(u32 fbWidth, u32 fbHeight) void SWRenderer::Swap(u32 fbWidth, u32 fbHeight)
{ {
GLInterface->Update(); // just updates the render window position and the backbuffer size GLInterface->Update(); // just updates the render window position and the backbuffer size
if (!g_SWVideoConfig.bHwRasterizer) SWRenderer::DrawTexture(GetCurrentColorTexture(), fbWidth, fbHeight);
SWRenderer::DrawTexture(GetCurrentColorTexture(), fbWidth, fbHeight);
swstats.frameCount++; swstats.frameCount++;
SWRenderer::SwapBuffer(); SWRenderer::SwapBuffer();

3
Source/Core/VideoBackends/Software/SWVideoConfig.cpp
View File

@ -14,7 +14,6 @@ SWVideoConfig::SWVideoConfig()
bHideCursor = false; bHideCursor = false;
renderToMainframe = false; renderToMainframe = false;
bHwRasterizer = false;
bBypassXFB = false; bBypassXFB = false;
bShowStats = false; bShowStats = false;
@ -42,7 +41,6 @@ void SWVideoConfig::Load(const char* ini_file)
hardware->Get("RenderToMainframe", &renderToMainframe, false); hardware->Get("RenderToMainframe", &renderToMainframe, false);
IniFile::Section* rendering = iniFile.GetOrCreateSection("Rendering"); IniFile::Section* rendering = iniFile.GetOrCreateSection("Rendering");
rendering->Get("HwRasterizer", &bHwRasterizer, false);
rendering->Get("BypassXFB", &bBypassXFB, false); rendering->Get("BypassXFB", &bBypassXFB, false);
rendering->Get("ZComploc", &bZComploc, true); rendering->Get("ZComploc", &bZComploc, true);
rendering->Get("ZFreeze", &bZFreeze, true); rendering->Get("ZFreeze", &bZFreeze, true);
@ -71,7 +69,6 @@ void SWVideoConfig::Save(const char* ini_file)
hardware->Set("RenderToMainframe", renderToMainframe); hardware->Set("RenderToMainframe", renderToMainframe);
IniFile::Section* rendering = iniFile.GetOrCreateSection("Rendering"); IniFile::Section* rendering = iniFile.GetOrCreateSection("Rendering");
rendering->Set("HwRasterizer", bHwRasterizer);
rendering->Set("BypassXFB", bBypassXFB); rendering->Set("BypassXFB", bBypassXFB);
rendering->Set("ZComploc", bZComploc); rendering->Set("ZComploc", bZComploc);
rendering->Set("ZFreeze", bZFreeze); rendering->Set("ZFreeze", bZFreeze);

1
Source/Core/VideoBackends/Software/SWVideoConfig.h
View File

@ -20,7 +20,6 @@ struct SWVideoConfig : NonCopyable
bool bHideCursor; bool bHideCursor;
bool renderToMainframe; bool renderToMainframe;
bool bHwRasterizer;
bool bBypassXFB; bool bBypassXFB;
// Emulation features // Emulation features

14
Source/Core/VideoBackends/Software/SWmain.cpp
View File

@ -17,12 +17,12 @@
#include "Core/HW/VideoInterface.h" #include "Core/HW/VideoInterface.h"
#include "VideoBackends/OGL/GLInterfaceBase.h" #include "VideoBackends/OGL/GLInterfaceBase.h"
#include "VideoBackends/OGL/GLUtil.h"
#include "VideoBackends/OGL/GLExtensions/GLExtensions.h" #include "VideoBackends/OGL/GLExtensions/GLExtensions.h"
#include "VideoBackends/Software/BPMemLoader.h" #include "VideoBackends/Software/BPMemLoader.h"
#include "VideoBackends/Software/Clipper.h" #include "VideoBackends/Software/Clipper.h"
#include "VideoBackends/Software/DebugUtil.h" #include "VideoBackends/Software/DebugUtil.h"
#include "VideoBackends/Software/EfbInterface.h" #include "VideoBackends/Software/EfbInterface.h"
#include "VideoBackends/Software/HwRasterizer.h"
#include "VideoBackends/Software/OpcodeDecoder.h" #include "VideoBackends/Software/OpcodeDecoder.h"
#include "VideoBackends/Software/Rasterizer.h" #include "VideoBackends/Software/Rasterizer.h"
#include "VideoBackends/Software/SWCommandProcessor.h" #include "VideoBackends/Software/SWCommandProcessor.h"
@ -91,7 +91,6 @@ bool VideoSoftware::Initialize(void *window_handle)
OpcodeDecoder::Init(); OpcodeDecoder::Init();
Clipper::Init(); Clipper::Init();
Rasterizer::Init(); Rasterizer::Init();
HwRasterizer::Init();
SWRenderer::Init(); SWRenderer::Init();
DebugUtil::Init(); DebugUtil::Init();
@ -155,7 +154,6 @@ void VideoSoftware::EmuStateChange(EMUSTATE_CHANGE newState)
void VideoSoftware::Shutdown() void VideoSoftware::Shutdown()
{ {
// TODO: should be in Video_Cleanup // TODO: should be in Video_Cleanup
HwRasterizer::Shutdown();
SWRenderer::Shutdown(); SWRenderer::Shutdown();
DebugUtil::Shutdown(); DebugUtil::Shutdown();
@ -190,7 +188,6 @@ void VideoSoftware::Video_Prepare()
// Do our OSD callbacks // Do our OSD callbacks
OSD::DoCallbacks(OSD::OSD_INIT); OSD::DoCallbacks(OSD::OSD_INIT);
HwRasterizer::Prepare();
SWRenderer::Prepare(); SWRenderer::Prepare();
INFO_LOG(VIDEO, "Video backend initialized."); INFO_LOG(VIDEO, "Video backend initialized.");
@ -221,14 +218,11 @@ void VideoSoftware::Video_EndField()
Core::Callback_VideoCopiedToXFB(false); Core::Callback_VideoCopiedToXFB(false);
return; return;
} }
if (!g_SWVideoConfig.bHwRasterizer) if (!g_SWVideoConfig.bBypassXFB)
{ {
if (!g_SWVideoConfig.bBypassXFB) EfbInterface::yuv422_packed *xfb = (EfbInterface::yuv422_packed *) Memory::GetPointer(s_beginFieldArgs.xfbAddr);
{
EfbInterface::yuv422_packed *xfb = (EfbInterface::yuv422_packed *) Memory::GetPointer(s_beginFieldArgs.xfbAddr);
SWRenderer::UpdateColorTexture(xfb, s_beginFieldArgs.fbWidth, s_beginFieldArgs.fbHeight); SWRenderer::UpdateColorTexture(xfb, s_beginFieldArgs.fbWidth, s_beginFieldArgs.fbHeight);
}
} }
// Ideally we would just move all the OpenGL context stuff to the CPU thread, // Ideally we would just move all the OpenGL context stuff to the CPU thread,

2
Source/Core/VideoBackends/Software/Software.vcxproj
View File

@ -41,7 +41,6 @@
<ClCompile Include="DebugUtil.cpp" /> <ClCompile Include="DebugUtil.cpp" />
<ClCompile Include="EfbCopy.cpp" /> <ClCompile Include="EfbCopy.cpp" />
<ClCompile Include="EfbInterface.cpp" /> <ClCompile Include="EfbInterface.cpp" />
<ClCompile Include="HwRasterizer.cpp" />
<ClCompile Include="OpcodeDecoder.cpp" /> <ClCompile Include="OpcodeDecoder.cpp" />
<ClCompile Include="RasterFont.cpp" /> <ClCompile Include="RasterFont.cpp" />
<ClCompile Include="Rasterizer.cpp" /> <ClCompile Include="Rasterizer.cpp" />
@ -65,7 +64,6 @@
<ClInclude Include="DebugUtil.h" /> <ClInclude Include="DebugUtil.h" />
<ClInclude Include="EfbCopy.h" /> <ClInclude Include="EfbCopy.h" />
<ClInclude Include="EfbInterface.h" /> <ClInclude Include="EfbInterface.h" />
<ClInclude Include="HwRasterizer.h" />
<ClInclude Include="NativeVertexFormat.h" /> <ClInclude Include="NativeVertexFormat.h" />
<ClInclude Include="OpcodeDecoder.h" /> <ClInclude Include="OpcodeDecoder.h" />
<ClInclude Include="RasterFont.h" /> <ClInclude Include="RasterFont.h" />

215
Source/Core/VideoBackends/Software/Tev.cpp
View File

@ -650,125 +650,120 @@ void Tev::Draw()
if (!TevAlphaTest(output[ALP_C])) if (!TevAlphaTest(output[ALP_C]))
return; return;
// This part is only needed if we are not simply computing bbox // z texture
// (i. e., only needed when using the SW renderer) if (bpmem.ztex2.op)
if (!BoundingBox::active)
{ {
// z texture u32 ztex = bpmem.ztex1.bias;
if (bpmem.ztex2.op) switch (bpmem.ztex2.type)
{ {
u32 ztex = bpmem.ztex1.bias; case 0: // 8 bit
switch (bpmem.ztex2.type) ztex += TexColor[ALP_C];
{ break;
case 0: // 8 bit case 1: // 16 bit
ztex += TexColor[ALP_C]; ztex += TexColor[ALP_C] << 8 | TexColor[RED_C];
break; break;
case 1: // 16 bit case 2: // 24 bit
ztex += TexColor[ALP_C] << 8 | TexColor[RED_C]; ztex += TexColor[RED_C] << 16 | TexColor[GRN_C] << 8 | TexColor[BLU_C];
break; break;
case 2: // 24 bit
ztex += TexColor[RED_C] << 16 | TexColor[GRN_C] << 8 | TexColor[BLU_C];
break;
}
if (bpmem.ztex2.op == ZTEXTURE_ADD)
ztex += Position[2];
Position[2] = ztex & 0x00ffffff;
} }
// fog if (bpmem.ztex2.op == ZTEXTURE_ADD)
if (bpmem.fog.c_proj_fsel.fsel) ztex += Position[2];
Position[2] = ztex & 0x00ffffff;
}
// fog
if (bpmem.fog.c_proj_fsel.fsel)
{
float ze;
if (bpmem.fog.c_proj_fsel.proj == 0)
{ {
float ze; // perspective
// ze = A/(B - (Zs >> B_SHF))
s32 denom = bpmem.fog.b_magnitude - (Position[2] >> bpmem.fog.b_shift);
//in addition downscale magnitude and zs to 0.24 bits
ze = (bpmem.fog.a.GetA() * 16777215.0f) / (float)denom;
}
else
{
// orthographic
// ze = a*Zs
//in addition downscale zs to 0.24 bits
ze = bpmem.fog.a.GetA() * ((float)Position[2] / 16777215.0f);
if (bpmem.fog.c_proj_fsel.proj == 0)
{
// perspective
// ze = A/(B - (Zs >> B_SHF))
s32 denom = bpmem.fog.b_magnitude - (Position[2] >> bpmem.fog.b_shift);
//in addition downscale magnitude and zs to 0.24 bits
ze = (bpmem.fog.a.GetA() * 16777215.0f) / (float)denom;
}
else
{
// orthographic
// ze = a*Zs
//in addition downscale zs to 0.24 bits
ze = bpmem.fog.a.GetA() * ((float)Position[2] / 16777215.0f);
}
if (bpmem.fogRange.Base.Enabled)
{
// TODO: This is untested and should definitely be checked against real hw.
// - No idea if offset is really normalized against the viewport width or against the projection matrix or yet something else
// - scaling of the "k" coefficient isn't clear either.
// First, calculate the offset from the viewport center (normalized to 0..1)
float offset = (Position[0] - (bpmem.fogRange.Base.Center - 342)) / (float)xfmem.viewport.wd;
// Based on that, choose the index such that points which are far away from the z-axis use the 10th "k" value and such that central points use the first value.
float floatindex = 9.f - std::abs(offset) * 9.f;
floatindex = (floatindex < 0.f) ? 0.f : (floatindex > 9.f) ? 9.f : floatindex; // TODO: This shouldn't be necessary!
// Get the two closest integer indices, look up the corresponding samples
int indexlower = (int)floor(floatindex);
int indexupper = indexlower + 1;
// Look up coefficient... Seems like multiplying by 4 makes Fortune Street work properly (fog is too strong without the factor)
float klower = bpmem.fogRange.K[indexlower/2].GetValue(indexlower%2) * 4.f;
float kupper = bpmem.fogRange.K[indexupper/2].GetValue(indexupper%2) * 4.f;
// linearly interpolate the samples and multiple ze by the resulting adjustment factor
float factor = indexupper - floatindex;
float k = klower * factor + kupper * (1.f - factor);
float x_adjust = sqrt(offset*offset + k*k)/k;
ze *= x_adjust; // NOTE: This is basically dividing by a cosine (hidden behind GXInitFogAdjTable): 1/cos = c/b = sqrt(a^2+b^2)/b
}
ze -= bpmem.fog.c_proj_fsel.GetC();
// clamp 0 to 1
float fog = (ze<0.0f) ? 0.0f : ((ze>1.0f) ? 1.0f : ze);
switch (bpmem.fog.c_proj_fsel.fsel)
{
case 4: // exp
fog = 1.0f - pow(2.0f, -8.0f * fog);
break;
case 5: // exp2
fog = 1.0f - pow(2.0f, -8.0f * fog * fog);
break;
case 6: // backward exp
fog = 1.0f - fog;
fog = pow(2.0f, -8.0f * fog);
break;
case 7: // backward exp2
fog = 1.0f - fog;
fog = pow(2.0f, -8.0f * fog * fog);
break;
}
// lerp from output to fog color
u32 fogInt = (u32)(fog * 256);
u32 invFog = 256 - fogInt;
output[RED_C] = (output[RED_C] * invFog + fogInt * bpmem.fog.color.r) >> 8;
output[GRN_C] = (output[GRN_C] * invFog + fogInt * bpmem.fog.color.g) >> 8;
output[BLU_C] = (output[BLU_C] * invFog + fogInt * bpmem.fog.color.b) >> 8;
} }
bool late_ztest = !bpmem.zcontrol.early_ztest || !g_SWVideoConfig.bZComploc; if (bpmem.fogRange.Base.Enabled)
if (late_ztest && bpmem.zmode.testenable)
{ {
// TODO: Check against hw if these values get incremented even if depth testing is disabled // TODO: This is untested and should definitely be checked against real hw.
EfbInterface::IncPerfCounterQuadCount(PQ_ZCOMP_INPUT); // - No idea if offset is really normalized against the viewport width or against the projection matrix or yet something else
// - scaling of the "k" coefficient isn't clear either.
if (!EfbInterface::ZCompare(Position[0], Position[1], Position[2])) // First, calculate the offset from the viewport center (normalized to 0..1)
return; float offset = (Position[0] - (bpmem.fogRange.Base.Center - 342)) / (float)xfmem.viewport.wd;
EfbInterface::IncPerfCounterQuadCount(PQ_ZCOMP_OUTPUT); // Based on that, choose the index such that points which are far away from the z-axis use the 10th "k" value and such that central points use the first value.
float floatindex = 9.f - std::abs(offset) * 9.f;
floatindex = (floatindex < 0.f) ? 0.f : (floatindex > 9.f) ? 9.f : floatindex; // TODO: This shouldn't be necessary!
// Get the two closest integer indices, look up the corresponding samples
int indexlower = (int)floor(floatindex);
int indexupper = indexlower + 1;
// Look up coefficient... Seems like multiplying by 4 makes Fortune Street work properly (fog is too strong without the factor)
float klower = bpmem.fogRange.K[indexlower/2].GetValue(indexlower%2) * 4.f;
float kupper = bpmem.fogRange.K[indexupper/2].GetValue(indexupper%2) * 4.f;
// linearly interpolate the samples and multiple ze by the resulting adjustment factor
float factor = indexupper - floatindex;
float k = klower * factor + kupper * (1.f - factor);
float x_adjust = sqrt(offset*offset + k*k)/k;
ze *= x_adjust; // NOTE: This is basically dividing by a cosine (hidden behind GXInitFogAdjTable): 1/cos = c/b = sqrt(a^2+b^2)/b
} }
ze -= bpmem.fog.c_proj_fsel.GetC();
// clamp 0 to 1
float fog = (ze<0.0f) ? 0.0f : ((ze>1.0f) ? 1.0f : ze);
switch (bpmem.fog.c_proj_fsel.fsel)
{
case 4: // exp
fog = 1.0f - pow(2.0f, -8.0f * fog);
break;
case 5: // exp2
fog = 1.0f - pow(2.0f, -8.0f * fog * fog);
break;
case 6: // backward exp
fog = 1.0f - fog;
fog = pow(2.0f, -8.0f * fog);
break;
case 7: // backward exp2
fog = 1.0f - fog;
fog = pow(2.0f, -8.0f * fog * fog);
break;
}
// lerp from output to fog color
u32 fogInt = (u32)(fog * 256);
u32 invFog = 256 - fogInt;
output[RED_C] = (output[RED_C] * invFog + fogInt * bpmem.fog.color.r) >> 8;
output[GRN_C] = (output[GRN_C] * invFog + fogInt * bpmem.fog.color.g) >> 8;
output[BLU_C] = (output[BLU_C] * invFog + fogInt * bpmem.fog.color.b) >> 8;
}
bool late_ztest = !bpmem.zcontrol.early_ztest || !g_SWVideoConfig.bZComploc;
if (late_ztest && bpmem.zmode.testenable)
{
// TODO: Check against hw if these values get incremented even if depth testing is disabled
EfbInterface::IncPerfCounterQuadCount(PQ_ZCOMP_INPUT);
if (!EfbInterface::ZCompare(Position[0], Position[1], Position[2]))
return;
EfbInterface::IncPerfCounterQuadCount(PQ_ZCOMP_OUTPUT);
} }
// branchless bounding box update // branchless bounding box update
@ -777,12 +772,6 @@ void Tev::Draw()
BoundingBox::coords[BoundingBox::TOP] = std::min((u16)Position[1], BoundingBox::coords[BoundingBox::TOP]); BoundingBox::coords[BoundingBox::TOP] = std::min((u16)Position[1], BoundingBox::coords[BoundingBox::TOP]);
BoundingBox::coords[BoundingBox::BOTTOM] = std::max((u16)Position[1], BoundingBox::coords[BoundingBox::BOTTOM]); BoundingBox::coords[BoundingBox::BOTTOM] = std::max((u16)Position[1], BoundingBox::coords[BoundingBox::BOTTOM]);
// if we are only calculating the bounding box,
// there's no need to actually draw anything
if (BoundingBox::active)
return;
#if ALLOW_TEV_DUMPS #if ALLOW_TEV_DUMPS
if (g_SWVideoConfig.bDumpTevStages) if (g_SWVideoConfig.bDumpTevStages)
{ {