Merge pull request #3070 from lioncash/vert

VertexShaderManager: Get rid of float pointer casts.
This commit is contained in:
flacs 2015-10-16 05:48:27 +02:00
commit 0d0704453e
3 changed files with 43 additions and 39 deletions

View File

@ -73,7 +73,7 @@ static void MultipleVec3Ortho(const Vec3 &vec, const float *proj, Vec4 &result)
void TransformPosition(const InputVertexData *src, OutputVertexData *dst)
{
const float* mat = (const float*)&xfmem.posMatrices[src->posMtx * 4];
const float* mat = &xfmem.posMatrices[src->posMtx * 4];
MultiplyVec3Mat34(src->position, mat, dst->mvPosition);
if (xfmem.projection.type == GX_PERSPECTIVE)
@ -88,7 +88,7 @@ void TransformPosition(const InputVertexData *src, OutputVertexData *dst)
void TransformNormal(const InputVertexData *src, bool nbt, OutputVertexData *dst)
{
const float* mat = (const float*)&xfmem.normalMatrices[(src->posMtx & 31) * 3];
const float* mat = &xfmem.normalMatrices[(src->posMtx & 31) * 3];
if (nbt)
{
@ -127,8 +127,8 @@ static void TransformTexCoordRegular(const TexMtxInfo &texinfo, int coordNum, bo
break;
}
const float *mat = (const float*)&xfmem.posMatrices[srcVertex->texMtx[coordNum] * 4];
Vec3 *dst = &dstVertex->texCoords[coordNum];
const float* mat = &xfmem.posMatrices[srcVertex->texMtx[coordNum] * 4];
Vec3* dst = &dstVertex->texCoords[coordNum];
if (texinfo.projection == XF_TEXPROJ_ST)
{
@ -153,7 +153,7 @@ static void TransformTexCoordRegular(const TexMtxInfo &texinfo, int coordNum, bo
// normalize
const PostMtxInfo &postInfo = xfmem.postMtxInfo[coordNum];
const float *postMat = (const float*)&xfmem.postMatrices[postInfo.index * 4];
const float* postMat = &xfmem.postMatrices[postInfo.index * 4];
if (specialCase)
{

View File

@ -4,9 +4,11 @@
#include <cfloat>
#include <cmath>
#include <cstring>
#include <sstream>
#include "Common/BitSet.h"
#include "Common/CommonFuncs.h"
#include "Common/CommonTypes.h"
#include "Common/MathUtil.h"
#include "Core/ConfigManager.h"
@ -236,7 +238,7 @@ void VertexShaderManager::SetConstants()
{
int startn = nTransformMatricesChanged[0] / 4;
int endn = (nTransformMatricesChanged[1] + 3) / 4;
memcpy(constants.transformmatrices[startn], &xfmem.posMatrices[startn * 4], (endn - startn) * 16);
memcpy(constants.transformmatrices[startn], &xfmem.posMatrices[startn * 4], (endn - startn) * sizeof(float4));
dirty = true;
nTransformMatricesChanged[0] = nTransformMatricesChanged[1] = -1;
}
@ -257,7 +259,7 @@ void VertexShaderManager::SetConstants()
{
int startn = nPostTransformMatricesChanged[0] / 4;
int endn = (nPostTransformMatricesChanged[1] + 3) / 4;
memcpy(constants.posttransformmatrices[startn], &xfmem.postMatrices[startn * 4], (endn - startn) * 16);
memcpy(constants.posttransformmatrices[startn], &xfmem.postMatrices[startn * 4], (endn - startn) * sizeof(float4));
dirty = true;
nPostTransformMatricesChanged[0] = nPostTransformMatricesChanged[1] = -1;
}
@ -331,30 +333,30 @@ void VertexShaderManager::SetConstants()
{
bPosNormalMatrixChanged = false;
const float *pos = (const float *)xfmem.posMatrices + g_main_cp_state.matrix_index_a.PosNormalMtxIdx * 4;
const float *norm = (const float *)xfmem.normalMatrices + 3 * (g_main_cp_state.matrix_index_a.PosNormalMtxIdx & 31);
const float* pos = &xfmem.posMatrices[g_main_cp_state.matrix_index_a.PosNormalMtxIdx * 4];
const float* norm = &xfmem.normalMatrices[3 * (g_main_cp_state.matrix_index_a.PosNormalMtxIdx & 31)];
memcpy(constants.posnormalmatrix, pos, 3*16);
memcpy(constants.posnormalmatrix[3], norm, 12);
memcpy(constants.posnormalmatrix[4], norm+3, 12);
memcpy(constants.posnormalmatrix[5], norm+6, 12);
memcpy(constants.posnormalmatrix, pos, 3 * sizeof(float4));
memcpy(constants.posnormalmatrix[3], norm, 3 * sizeof(float));
memcpy(constants.posnormalmatrix[4], norm + 3, 3 * sizeof(float));
memcpy(constants.posnormalmatrix[5], norm + 6, 3 * sizeof(float));
dirty = true;
}
if (bTexMatricesChanged[0])
{
bTexMatricesChanged[0] = false;
const float *fptrs[] =
const float* pos_matrix_ptrs[] =
{
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex0MtxIdx * 4],
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex1MtxIdx * 4],
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex2MtxIdx * 4],
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex3MtxIdx * 4]
&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex0MtxIdx * 4],
&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex1MtxIdx * 4],
&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex2MtxIdx * 4],
&xfmem.posMatrices[g_main_cp_state.matrix_index_a.Tex3MtxIdx * 4]
};
for (int i = 0; i < 4; ++i)
for (size_t i = 0; i < ArraySize(pos_matrix_ptrs); ++i)
{
memcpy(constants.texmatrices[3 * i], fptrs[i], 3 * 16);
memcpy(constants.texmatrices[3 * i], pos_matrix_ptrs[i], 3 * sizeof(float4));
}
dirty = true;
}
@ -362,16 +364,16 @@ void VertexShaderManager::SetConstants()
if (bTexMatricesChanged[1])
{
bTexMatricesChanged[1] = false;
const float *fptrs[] = {
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex4MtxIdx * 4],
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex5MtxIdx * 4],
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex6MtxIdx * 4],
(const float *)&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex7MtxIdx * 4]
const float* pos_matrix_ptrs[] = {
&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex4MtxIdx * 4],
&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex5MtxIdx * 4],
&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex6MtxIdx * 4],
&xfmem.posMatrices[g_main_cp_state.matrix_index_b.Tex7MtxIdx * 4]
};
for (int i = 0; i < 4; ++i)
for (size_t i = 0; i < ArraySize(pos_matrix_ptrs); ++i)
{
memcpy(constants.texmatrices[3*i + 12], fptrs[i], 3*16);
memcpy(constants.texmatrices[3*i + 12], pos_matrix_ptrs[i], 3 * sizeof(float4));
}
dirty = true;
}
@ -531,7 +533,7 @@ void VertexShaderManager::SetConstants()
Matrix44::Set(mtxB, g_fProjectionMatrix);
Matrix44::Multiply(mtxB, viewMtx, mtxA); // mtxA = projection x view
Matrix44::Multiply(s_viewportCorrection, mtxA, mtxB); // mtxB = viewportCorrection x mtxA
memcpy(constants.projection, mtxB.data, 4*16);
memcpy(constants.projection, mtxB.data, 4 * sizeof(float4));
}
else
{
@ -540,7 +542,7 @@ void VertexShaderManager::SetConstants()
Matrix44 correctedMtx;
Matrix44::Multiply(s_viewportCorrection, projMtx, correctedMtx);
memcpy(constants.projection, correctedMtx.data, 4*16);
memcpy(constants.projection, correctedMtx.data, 4 * sizeof(float4));
}
dirty = true;
@ -695,7 +697,7 @@ void VertexShaderManager::TranslateView(float x, float y, float z)
Matrix33::Multiply(s_viewInvRotationMatrix, vector, result);
for (int i = 0; i < 3; i++)
for (size_t i = 0; i < ArraySize(result); i++)
s_fViewTranslationVector[i] += result[i];
bProjectionChanged = true;
@ -732,16 +734,18 @@ void VertexShaderManager::ResetView()
void VertexShaderManager::TransformToClipSpace(const float* data, float* out, u32 MtxIdx)
{
const float* world_matrix = (const float*)xfmem.posMatrices + (MtxIdx & 0x3f) * 4;
// We use the projection matrix calculated by vertexShaderManager, because it
const float* world_matrix = &xfmem.posMatrices[(MtxIdx & 0x3f) * 4];
// We use the projection matrix calculated by VertexShaderManager, because it
// includes any free look transformations.
// Make sure VertexManager::SetConstants() has been called first.
const float* proj_matrix = &g_fProjectionMatrix[0];
float t[3];
t[0] = data[0] * world_matrix[0] + data[1] * world_matrix[1] + data[2] * world_matrix[2] + world_matrix[3];
t[1] = data[0] * world_matrix[4] + data[1] * world_matrix[5] + data[2] * world_matrix[6] + world_matrix[7];
t[2] = data[0] * world_matrix[8] + data[1] * world_matrix[9] + data[2] * world_matrix[10] + world_matrix[11];
const float t[3] = {
data[0] * world_matrix[0] + data[1] * world_matrix[1] + data[2] * world_matrix[2] + world_matrix[3],
data[0] * world_matrix[4] + data[1] * world_matrix[5] + data[2] * world_matrix[6] + world_matrix[7],
data[0] * world_matrix[8] + data[1] * world_matrix[9] + data[2] * world_matrix[10] + world_matrix[11]
};
out[0] = t[0] * proj_matrix[0] + t[1] * proj_matrix[1] + t[2] * proj_matrix[2] + proj_matrix[3];
out[1] = t[0] * proj_matrix[4] + t[1] * proj_matrix[5] + t[2] * proj_matrix[6] + proj_matrix[7];

View File

@ -267,11 +267,11 @@ struct Projection
struct XFMemory
{
u32 posMatrices[256]; // 0x0000 - 0x00ff
float posMatrices[256]; // 0x0000 - 0x00ff
u32 unk0[768]; // 0x0100 - 0x03ff
u32 normalMatrices[96]; // 0x0400 - 0x045f
float normalMatrices[96]; // 0x0400 - 0x045f
u32 unk1[160]; // 0x0460 - 0x04ff
u32 postMatrices[256]; // 0x0500 - 0x05ff
float postMatrices[256]; // 0x0500 - 0x05ff
Light lights[8]; // 0x0600 - 0x067f
u32 unk2[2432]; // 0x0680 - 0x0fff
u32 error; // 0x1000