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

Moved components' perks computation to the DecodedVArray in DX9. Performance boost. Working on OGL version 

git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@771 8ced0084-cf51-0410-be5f-012b33b47a6e
This commit is contained in:
XTra.KrazzY 2008-10-04 21:47:56 +00:00
parent 664f4dee0e
commit 8eef6f9cd5
3 changed files with 40 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
Source/Plugins/Plugin_VideoDX9/Src/DecodedVArray.h
View File

@ -46,7 +46,8 @@ public:
int count;
DecodedVArray();
~DecodedVArray();
void SetComponents(u32 comps) {components = comps; vertexSize = ComputeVertexSize(components);}
void SetComponents(u32 comps) {components = comps; vertexSize = ComputeVertexSize(components);
ComputeComponents(); }
u32 GetComponents() const {return components;}
void Create(int _size, int pmcount, int tmcount, int nrmcount, int colcount, int tccount);
void Zero();
@ -84,6 +85,14 @@ public:
uvs[n][count].u=u;
uvs[n][count].v=v;
}
void ComputeComponents() {
hasPosMatIdx = (components & (1 << 1)) != 0;
for(int i = 0; i < 8; i++)
hasTexMatIdx[i] = (components & (1 << (i + 2))) != 0;
hasNrm = (components & (1 << 10)) != 0;
}
const DecPos &GetPos(int n) const { return positions[n]; }
const DecNormal &GetNormal(int i, int n) const { return normals[i][n]; }
const DecColor &GetColor(int i, int n) const { return colors[i][n]; }
@ -97,6 +106,9 @@ public:
DecUV *uvs[8];
DecMtxInd *posMtxInds;
DecMtxInd *texMtxInds[8];
// Component data
bool hasPosMatIdx, hasTexMatIdx[8], hasNrm;
};
#endif

37
Source/Plugins/Plugin_VideoDX9/Src/TransformEngine.cpp
View File

@ -15,6 +15,8 @@
// Official SVN repository and contact information can be found at
// http://code.google.com/p/dolphin-emu/
#include <fvec.h>
#include "Common.h"
#include "Globals.h"
#include "Vec3.h"
@ -82,18 +84,31 @@ float DoLighting(const Light *light, const LitChannel &chan, const Vec3 &pos, co
return val;
}
void VtxMulMtx43(Vec3 &out, const Vec3 &in, const float pMatrix[12])
void VtxMulMtx43T(Vec3 &out, const Vec3 &in, const float pMatrix[12])
{
out.x = in.x * pMatrix[0] + in.y * pMatrix[1] + in.z * pMatrix[2] + 1 * pMatrix[3];
out.y = in.x * pMatrix[4] + in.y * pMatrix[5] + in.z * pMatrix[6] + 1 * pMatrix[7];
out.z = in.x * pMatrix[8] + in.y * pMatrix[9] + in.z * pMatrix[10] + 1 * pMatrix[11];
}
void VtxMulMtx43T(Vec3 &out, const Vec3 &in, const float pMatrix[12])
void VtxMulMtx43(Vec3 &out, const Vec3 &in, const float pMatrix[12])
{
out.x = in.x * pMatrix[0] + in.y * pMatrix[1] + in.z * pMatrix[2] + 1 * pMatrix[3];
out.y = in.x * pMatrix[4] + in.y * pMatrix[5] + in.z * pMatrix[6] + 1 * pMatrix[7];
out.z = in.x * pMatrix[8] + in.y * pMatrix[9] + in.z * pMatrix[10] + 1 * pMatrix[11];
VtxMulMtx43T(out,in,pMatrix);
//TODO(XK): Turns out that SSE2 computations are slower... Can anyone do
// anything about it?
/*
F32vec4 a(in.x, in.y, in.z, 1), b(pMatrix[0], pMatrix[1], pMatrix[2], pMatrix[3]);
out.x = add_horizontal(a * b);
b[0] = pMatrix[4]; b[1] = pMatrix[5]; b[2] = pMatrix[6]; b[3] = pMatrix[7];
out.y = add_horizontal(a * b);
b[0] = pMatrix[8]; b[1] = pMatrix[9]; b[2] = pMatrix[10]; b[3] = pMatrix[11];
out.z = add_horizontal(a * b);
*/
}
void VtxMulMtx42(Vec3 &out, const Vec3 &in, const float pMatrix[8])
@ -122,6 +137,8 @@ void CTransformEngine::TransformVertices(int _numVertices, const DecodedVArray *
RGBAFloat lightVals[8];
RGBAFloat chans[2];
u32 components = varray->GetComponents();
// TODO: only for active lights
for (int i=0; i<8; i++)
lightColors[i].convert_GC(GetLight(i)->color);
@ -134,7 +151,7 @@ void CTransformEngine::TransformVertices(int _numVertices, const DecodedVArray *
Vec3 OrigPos = varray->GetPos(i);
if (varray->GetComponents() & VertexLoader::VB_HAS_POSMTXIDX)
if (varray->hasPosMatIdx)
{
int index = varray->GetPosMtxInd(i);
SetPosNormalMatrix(
@ -144,7 +161,7 @@ void CTransformEngine::TransformVertices(int _numVertices, const DecodedVArray *
for (int j = 0; j < 8; j++)
{
if (varray->GetComponents() & (VertexLoader::VB_HAS_TEXMTXIDX0<<j))
if (varray->hasTexMatIdx[j])
{
float *flipmem = (float *)xfmem;
int index = varray->GetTexMtxInd(j, i);
@ -159,7 +176,7 @@ void CTransformEngine::TransformVertices(int _numVertices, const DecodedVArray *
Vec3 TempNormal;
Vec3 OrigNormal;
if (varray->GetComponents() & VertexLoader::VB_HAS_NRM0)
if (varray->hasNrm)
{
OrigNormal = varray->GetNormal(0, i);
VtxMulMtx33(TempNormal, OrigNormal, m_pNormalMatrix);
@ -188,7 +205,7 @@ void CTransformEngine::TransformVertices(int _numVertices, const DecodedVArray *
RGBAFloat material;
RGBAFloat lightSum(0,0,0,0);
bool hasColorJ = (varray->GetComponents() & (VertexLoader::VB_HAS_COL0 << j)) != 0;
bool hasColorJ = (components & (VertexLoader::VB_HAS_COL0 << j)) != 0;
//get basic material color from appropriate sources (this would compile nicely!:)
if (xfregs.colChans[j].color.matsource == GX_SRC_REG)
@ -303,7 +320,7 @@ void CTransformEngine::TransformVertices(int _numVertices, const DecodedVArray *
default:
{
int c = xfregs.texcoords[n].texmtxinfo.sourcerow - XF_SRCTEX0_INROW;
bool hasTCC = (varray->GetComponents() & (VertexLoader::VB_HAS_UV0 << c)) != 0;
bool hasTCC = (components & (VertexLoader::VB_HAS_UV0 << c)) != 0;
if (c >= 0 && c <= 7 && hasTCC)
{
const DecUV &uv = varray->GetUV(c, i);

1
Source/Plugins/Plugin_VideoDX9/Src/VertexLoader.cpp
View File

@ -366,7 +366,6 @@ void VertexLoader::PrepareRun()
for (int i = 0; i < 2; i++)
colElements[i] = m_VtxAttr.color[i].Elements;
varray->Reset();
varray->SetComponents(m_components);
}