MGislv/GLideN64
1
0
Fork 0
mirror of https://github.com/blawar/GLideN64.git synced 2024-07-02 09:03:37 +00:00
Code Issues Projects Releases Wiki Activity

Rewrite RDRAMtoFrameBuffer::CopyFromRDRAM:

Browse Source 
copy from RDRAM only those pixels, which were provided with FBWrite.
This commit is contained in:
Sergey Lipskiy 2016-02-03 21:17:53 +06:00
parent dcf32d0c5d
commit ce7526eaf5
4 changed files with 157 additions and 110 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

253
src/FrameBuffer.cpp
View File

@ -1,6 +1,7 @@
#include <algorithm>
#include <assert.h>
#include <math.h>
#include <algorithm>
#include <vector>
#include "OpenGL.h"
#include "FrameBuffer.h"
#include "DepthBuffer.h"
@ -104,10 +105,7 @@ class RDRAMtoFrameBuffer
{
public:
RDRAMtoFrameBuffer()
: m_uly(0)
, m_lry(0)
, m_stride(0)
, m_pCurBuffer(nullptr)
: m_pCurBuffer(nullptr)
, m_pTexture(nullptr)
, m_PBO(0) {
}
@ -115,32 +113,25 @@ public:
void Init();
void Destroy();
void AddAddress(u32 _address);
void CopyFromRDRAM(u32 _address, bool _bUseAlpha);
void AddAddress(u32 _address, u32 _size);
void CopyFromRDRAM(u32 _address, bool _bCFB);
private:
class Cleaner {
public:
Cleaner(RDRAMtoFrameBuffer * _p) : m_p(_p) {}
~Cleaner() {
m_p->m_uly = 0;
m_p->m_lry = 0;
m_p->m_stride = 0;
m_p->m_pCurBuffer = nullptr;
m_p->m_vecAddress.clear();
}
private:
RDRAMtoFrameBuffer * m_p;
};
u32 m_uly, m_lry;
u32 m_stride;
FrameBuffer * m_pCurBuffer;
CachedTexture * m_pTexture;
#ifndef GLES2
GLuint m_PBO;
#else
GLubyte* m_PBO;
#endif
std::vector<u32> m_vecAddress;
};
#ifndef GLES2
@ -1548,116 +1539,176 @@ void RDRAMtoFrameBuffer::Destroy()
#endif
}
void RDRAMtoFrameBuffer::AddAddress(u32 _address)
void RDRAMtoFrameBuffer::AddAddress(u32 _address, u32 _size)
{
if (m_pCurBuffer == nullptr) {
m_pCurBuffer = frameBufferList().findBuffer(_address);
if (m_pCurBuffer == nullptr)
return;
m_stride = m_pCurBuffer->m_width << m_pCurBuffer->m_size >> 1;
m_uly = m_lry = (_address - m_pCurBuffer->m_startAddress) / m_stride;
return;
}
const u32 y = (_address - m_pCurBuffer->m_startAddress) / m_stride;
if (y < m_uly)
m_uly = y;
else if (y > m_lry)
m_lry = y;
const u32 pixelSize = 1 << m_pCurBuffer->m_size >> 1;
if (_size != pixelSize && (_address%pixelSize) > 0)
return;
m_vecAddress.push_back(_address);
gDP.colorImage.changed = TRUE;
}
void RDRAMtoFrameBuffer::CopyFromRDRAM( u32 _address, bool _bUseAlpha)
// Write the whole buffer
template <typename TSrc>
bool _copyBufferFromRdram(u32 _address, u32* _dst, u32(*converter)(TSrc _c, bool _bCFB), u32 _xor, u32 _x0, u32 _y0, u32 _width, u32 _height, bool _bCFB)
{
FrameBuffer *pBuffer = frameBufferList().findBuffer(_address);
if (pBuffer == NULL || pBuffer->m_size < G_IM_SIZ_16b)
return;
if (pBuffer->m_startAddress == _address && gDP.colorImage.changed != 0)
TSrc * src = reinterpret_cast<TSrc*>(RDRAM + _address);
const u32 bound = (RDRAMSize + 1 - _address) >> (sizeof(TSrc) / 2);
TSrc col;
u32 idx;
u32 summ = 0;
u32 dsty = 0;
const u32 y1 = _y0 + _height;
for (u32 y = _y0; y < y1; ++y) {
for (u32 x = _x0; x < _width; ++x) {
idx = (x + (_height - y - 1)*_width) ^ _xor;
if (idx >= bound)
break;
col = src[idx];
summ += col;
_dst[x + dsty*_width] = converter(col, _bCFB);
}
++dsty;
}
return summ != 0;
}
// Write only pixels provided with FBWrite
template <typename TSrc>
bool _copyPixelsFromRdram(u32 _address, const vector<u32> & _vecAddress, u32* _dst, u32(*converter)(TSrc _c, bool _bCFB), u32 _xor, u32 _width, u32 _height, bool _bCFB)
{
memset(_dst, 0, _width*_height*sizeof(u32));
TSrc * src = reinterpret_cast<TSrc*>(RDRAM + _address);
const u32 szPixel = sizeof(TSrc);
const size_t numPixels = _vecAddress.size();
TSrc col;
u32 summ = 0;
u32 idx, w, h;
for (size_t i = 0; i < numPixels; ++i) {
idx = (_vecAddress[i] - _address) / szPixel;
w = idx % _width;
h = idx / _width;
col = src[idx];
summ += col;
_dst[(w + (_height - h)*_width) ^ _xor] = converter(col, _bCFB);
}
return summ != 0;
}
u32 RGBA16ToABGR32(u16 col, bool _bCFB)
{
u32 r, g, b, a;
r = ((col >> 11) & 31) << 3;
g = ((col >> 6) & 31) << 3;
b = ((col >> 1) & 31) << 3;
if (_bCFB)
a = 0xFF;
else
a = (col & 1) > 0 && (r | g | b) > 0 ? 0xFF : 0U;
return ((a << 24) | (b << 16) | (g << 8) | r);
}
u32 RGBA32ToABGR32(u32 col, bool _bCFB)
{
u32 r, g, b, a;
r = (col >> 24) & 0xff;
g = (col >> 16) & 0xff;
b = (col >> 8) & 0xff;
if (_bCFB)
a = 0xFF;
else
a = (r | g | b) > 0 ? col & 0xFF : 0U;
return ((a << 24) | (b << 16) | (g << 8) | r);
}
void RDRAMtoFrameBuffer::CopyFromRDRAM(u32 _address, bool _bCFB)
{
Cleaner cleaner(this);
if (m_pCurBuffer == nullptr) {
if (!_bCFB && config.frameBufferEmulation.copyFromRDRAM == 0)
return;
m_pCurBuffer = frameBufferList().findBuffer(_address);
}
else if (m_vecAddress.empty())
return;
const bool bUseAlpha = _bUseAlpha && pBuffer->m_changed;
const u32 address = pBuffer->m_startAddress;
const u32 width = pBuffer->m_width;
const u32 height = _cutHeight(address, pBuffer->m_startAddress == _address ? VI.real_height : pBuffer->m_height, pBuffer->m_width * 2);
if (m_pCurBuffer == nullptr || m_pCurBuffer->m_size < G_IM_SIZ_16b)
return;
if (m_pCurBuffer->m_startAddress == _address && gDP.colorImage.changed != 0)
return;
const u32 address = m_pCurBuffer->m_startAddress;
const u32 height = _cutHeight(address, m_pCurBuffer->m_startAddress == _address ? VI.real_height : m_pCurBuffer->m_height, m_pCurBuffer->m_width << m_pCurBuffer->m_size >> 1);
if (height == 0)
return;
const u32 width = m_pCurBuffer->m_width;
const u32 x0 = 0;
const u32 y0 = 0;
const u32 y1 = y0 + height;
const bool bUseAlpha = !_bCFB && m_pCurBuffer->m_changed;
m_pTexture->width = width;
m_pTexture->height = height;
const u32 dataSize = width*height*4;
const u32 dataSize = width*height * 4;
#ifndef GLES2
PBOBinder binder(GL_PIXEL_UNPACK_BUFFER, m_PBO);
glBufferData(GL_PIXEL_UNPACK_BUFFER, dataSize, NULL, GL_DYNAMIC_DRAW);
GLubyte* ptr = (GLubyte*)glMapBufferRange(GL_PIXEL_UNPACK_BUFFER, 0, dataSize, GL_MAP_WRITE_BIT);
#else
m_PBO = (GLubyte*)malloc(dataSize);
GLubyte* ptr = m_PBO;
PBOBinder binder(m_PBO);
GLubyte* ptr = (GLubyte*)malloc(dataSize);
PBOBinder binder(ptr);
#endif // GLES2
if (ptr == NULL)
return;
u8 * image = RDRAM + address;
u32 * dst = (u32*)ptr;
u32 empty = 0;
u32 r, g, b, a, idx;
if (pBuffer->m_size == G_IM_SIZ_16b) {
u16 * src = (u16*)image;
u16 col;
const u32 bound = (RDRAMSize + 1 - address) >> 1;
for (u32 y = 0; y < height; y++)
{
for (u32 x = 0; x < width; x++)
{
idx = (x + (height - y - 1)*width)^1;
if (idx >= bound)
break;
col = src[idx];
if (bUseAlpha)
src[idx] = 0;
empty |= col;
r = ((col >> 11)&31)<<3;
g = ((col >> 6)&31)<<3;
b = ((col >> 1)&31)<<3;
a = (col&1) > 0 && (r|g|b) > 0 ? 0xff : 0U;
dst[x + y*width] = (a << 24) | (b << 16) | (g << 8) | r;
}
}
} else {
// 32 bit
u32 * src = (u32*)image;
u32 col;
const u32 bound = (RDRAMSize + 1 - address) >> 2;
for (u32 y=0; y < height; y++)
{
for (u32 x=0; x < width; x++)
{
idx = x + (height - y - 1)*width;
if (idx >= bound)
break;
col = src[idx];
if (bUseAlpha)
src[idx] = 0;
empty |= col;
r = (col >> 24) & 0xff;
g = (col >> 16) & 0xff;
b = (col >> 8) & 0xff;
a = (r|g|b) > 0 ? col & 0xff : 0U;
dst[x + y*width] = (a<<24)|(b<<16)|(g<<8)|r;
}
}
bool bCopy;
if (m_vecAddress.empty()) {
if (m_pCurBuffer->m_size == G_IM_SIZ_16b)
bCopy = _copyBufferFromRdram<u16>(address, dst, RGBA16ToABGR32, 1, x0, y0, width, height, _bCFB);
else
bCopy = _copyBufferFromRdram<u32>(address, dst, RGBA32ToABGR32, 0, x0, y0, width, height, _bCFB);
}
else {
if (m_pCurBuffer->m_size == G_IM_SIZ_16b)
bCopy = _copyPixelsFromRdram<u16>(address, m_vecAddress, dst, RGBA16ToABGR32, 1, width, height, _bCFB);
else
bCopy = _copyPixelsFromRdram<u32>(address, m_vecAddress, dst, RGBA32ToABGR32, 0, width, height, _bCFB);
}
if (bUseAlpha) {
u32 totalBytes = (width * height) << m_pCurBuffer->m_size >> 1;
if (address + totalBytes > RDRAMSize + 1)
totalBytes = RDRAMSize + 1 - address;
memset(RDRAM + address, 0, totalBytes);
}
#ifndef GLES2
glUnmapBuffer(GL_PIXEL_UNPACK_BUFFER); // release the mapped buffer
#endif
if (empty == 0)
if (!bCopy)
return;
glBindTexture(GL_TEXTURE_2D, m_pTexture->glName);
#ifndef GLES2
glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, GL_RGBA, GL_UNSIGNED_BYTE, 0);
#else
glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, GL_RGBA, GL_UNSIGNED_BYTE, m_PBO);
glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, GL_RGBA, GL_UNSIGNED_BYTE, ptr);
free(ptr);
#endif
m_pTexture->scaleS = 1.0f / (float)m_pTexture->realWidth;
@ -1673,23 +1724,17 @@ void RDRAMtoFrameBuffer::CopyFromRDRAM( u32 _address, bool _bUseAlpha)
gDPTile * pTile0 = gSP.textureTile[0];
gSP.textureTile[0] = &tile0;
if (_bUseAlpha) {
CombinerInfo::get().setCombine(EncodeCombineMode(0, 0, 0, TEXEL0, 0, 0, 0, TEXEL0, 0, 0, 0, TEXEL0, 0, 0, 0, TEXEL0));
glEnable(GL_BLEND);
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
}
else {
CombinerInfo::get().setCombine(EncodeCombineMode(0, 0, 0, TEXEL0, 0, 0, 0, 1, 0, 0, 0, TEXEL0, 0, 0, 0, 1));
glDisable(GL_BLEND);
}
CombinerInfo::get().setCombine(EncodeCombineMode(0, 0, 0, TEXEL0, 0, 0, 0, TEXEL0, 0, 0, 0, TEXEL0, 0, 0, 0, TEXEL0));
glEnable(GL_BLEND);
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
currentCombiner()->updateFrameBufferInfo();
glDisable(GL_DEPTH_TEST);
const u32 gdpChanged = gDP.changed & CHANGED_CPU_FB_WRITE;
gSP.changed = gDP.changed = 0;
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, pBuffer->m_FBO);
OGLRender::TexturedRectParams params(0.0f, 0.0f, (float)width, (float)height, 0.0f, 0.0f, width - 1.0f, height - 1.0f, false);
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, m_pCurBuffer->m_FBO);
OGLRender::TexturedRectParams params((float)x0, (float)y0, (float)width, (float)height, 0.0f, 0.0f, width - 1.0f, height - 1.0f, false);
video().getRender().drawTexturedRect(params);
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, frameBufferList().getCurrent()->m_FBO);
@ -1698,12 +1743,12 @@ void RDRAMtoFrameBuffer::CopyFromRDRAM( u32 _address, bool _bUseAlpha)
gDP.changed |= gdpChanged | CHANGED_RENDERMODE | CHANGED_COMBINE;
}
void FrameBuffer_CopyFromRDRAM( u32 address, bool bUseAlpha )
void FrameBuffer_CopyFromRDRAM(u32 _address, bool _bCFB)
{
g_RDRAMtoFB.CopyFromRDRAM(address, bUseAlpha);
g_RDRAMtoFB.CopyFromRDRAM(_address, _bCFB);
}
void FrameBuffer_AddAddress(u32 address)
void FrameBuffer_AddAddress(u32 address, u32 _size)
{
g_RDRAMtoFB.AddAddress(address);
g_RDRAMtoFB.AddAddress(address, _size);
}

2
src/FrameBuffer.h
View File

@ -121,7 +121,7 @@ void FrameBuffer_Destroy();
void FrameBuffer_CopyToRDRAM( u32 _address , bool _sync );
void FrameBuffer_CopyChunkToRDRAM(u32 _address);
void FrameBuffer_CopyFromRDRAM(u32 address, bool bUseAlpha);
void FrameBuffer_AddAddress(u32 address);
void FrameBuffer_AddAddress(u32 address, u32 _size);
bool FrameBuffer_CopyDepthBuffer(u32 address);
bool FrameBuffer_CopyDepthBufferChunk(u32 address);
void FrameBuffer_ActivateBufferTexture(s16 t, FrameBuffer *pBuffer);

7
src/FrameBufferInfo.cpp
View File

@ -13,11 +13,12 @@ FBInfo fbInfo;
void FBInfo::Write(u32 addr, u32 size)
{
// TODO: remove debug print
debugPrint("FBWrite addr=%08lx size=%u\n", addr, size);
//debugPrint("FBWrite addr=%08lx size=%u\n", addr, size);
const u32 address = RSP_SegmentToPhysical(addr);
m_pWriteBuffer = frameBufferList().findBuffer(address);
FrameBuffer_AddAddress(address);
if (m_pWriteBuffer == nullptr)
m_pWriteBuffer = frameBufferList().findBuffer(address);
FrameBuffer_AddAddress(address, size);
}
void FBInfo::WriteList(FrameBufferModifyEntry *plist, u32 size)

5
src/VI.cpp
View File

@ -144,12 +144,13 @@ void VI_UpdateScreen()
frameBufferList().saveBuffer(*REG.VI_ORIGIN, G_IM_FMT_RGBA, size, VI.width, VI.height, true);
}
}
if ((((*REG.VI_STATUS) & 3) > 0) && ((config.frameBufferEmulation.copyFromRDRAM && gDP.colorImage.changed) || bCFB)) {
// if ((((*REG.VI_STATUS) & 3) > 0) && (gDP.colorImage.changed || bCFB)) { // Does not work in release build!!!
if (((*REG.VI_STATUS) & 3) > 0) {
if (!bVIUpdated) {
VI_UpdateSize();
bVIUpdated = true;
}
FrameBuffer_CopyFromRDRAM(*REG.VI_ORIGIN, config.frameBufferEmulation.copyFromRDRAM && !bCFB);
FrameBuffer_CopyFromRDRAM(*REG.VI_ORIGIN, bCFB);
}
frameBufferList().renderBuffer(*REG.VI_ORIGIN);
frameBufferList().clearBuffersChanged();