Merge remote-tracking branch 'remotes/origin/dx9-ssaa-fix'

This commit is contained in:
Rodolfo Bogado 2013-08-23 22:43:03 -03:00
commit dbcc677922
8 changed files with 252 additions and 198 deletions

View file

@ -97,6 +97,7 @@ void VideoConfig::Load(const char *ini_file)
iniFile.Get("Hacks", "EFBScaledCopy", &bCopyEFBScaled, true);
iniFile.Get("Hacks", "EFBCopyCacheEnable", &bEFBCopyCacheEnable, false);
iniFile.Get("Hacks", "EFBEmulateFormatChanges", &bEFBEmulateFormatChanges, false);
iniFile.Get("Hacks", "ForceDualSourceBlend", &bForceDualSourceBlend, false);
iniFile.Get("Hardware", "Adapter", &iAdapter, 0);
@ -265,6 +266,7 @@ void VideoConfig::Save(const char *ini_file)
iniFile.Set("Hacks", "EFBScaledCopy", bCopyEFBScaled);
iniFile.Set("Hacks", "EFBCopyCacheEnable", bEFBCopyCacheEnable);
iniFile.Set("Hacks", "EFBEmulateFormatChanges", bEFBEmulateFormatChanges);
iniFile.Set("Hacks", "ForceDualSourceBlend", bForceDualSourceBlend);
iniFile.Set("Hardware", "Adapter", iAdapter);

View file

@ -124,7 +124,8 @@ struct VideoConfig
bool bEnablePixelLighting;
bool bHackedBufferUpload;
bool bFastDepthCalc;
//for dx9-backend
bool bForceDualSourceBlend;
int iLog; // CONF_ bits
int iSaveTargetId; // TODO: Should be dropped

View file

@ -35,6 +35,8 @@ void PerfQuery::DestroyDeviceObjects()
void PerfQuery::EnableQuery(PerfQueryGroup type)
{
if (!ShouldEmulate())
return;
// Is this sane?
if (m_query_count > ARRAYSIZE(m_query_buffer) / 2)
WeakFlush();
@ -58,6 +60,8 @@ void PerfQuery::EnableQuery(PerfQueryGroup type)
void PerfQuery::DisableQuery(PerfQueryGroup type)
{
if (!ShouldEmulate())
return;
// stop query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
@ -74,6 +78,8 @@ void PerfQuery::ResetQuery()
u32 PerfQuery::GetQueryResult(PerfQueryType type)
{
if (!ShouldEmulate())
return 0;
u32 result = 0;
if (type == PQ_ZCOMP_INPUT_ZCOMPLOC || type == PQ_ZCOMP_OUTPUT_ZCOMPLOC)
@ -98,6 +104,8 @@ u32 PerfQuery::GetQueryResult(PerfQueryType type)
void PerfQuery::FlushOne()
{
if (!ShouldEmulate())
return;
auto& entry = m_query_buffer[m_query_read_pos];
DWORD result = 0;
@ -118,12 +126,16 @@ void PerfQuery::FlushOne()
// TODO: could selectively flush things, but I don't think that will do much
void PerfQuery::FlushResults()
{
if (!ShouldEmulate())
return;
while (!IsFlushed())
FlushOne();
}
void PerfQuery::WeakFlush()
{
if (!ShouldEmulate())
return;
while (!IsFlushed())
{
auto& entry = m_query_buffer[m_query_read_pos];
@ -148,6 +160,8 @@ void PerfQuery::WeakFlush()
bool PerfQuery::IsFlushed() const
{
if (!ShouldEmulate())
return true;
return 0 == m_query_count;
}

View file

@ -117,30 +117,30 @@ LPDIRECT3DPIXELSHADER9 PixelShaderCache::ReinterpRGB8ToRGBA6()
/* old code here for reference
const char code[] =
{
"uniform sampler samp0 : register(s0);\n"
"void main(\n"
" out float4 ocol0 : COLOR0,\n"
" in float2 uv0 : TEXCOORD0){\n"
" ocol0 = tex2D(samp0,uv0);\n"
" float4 src8 = round(ocol0*255.f);\n"
" ocol0.r = floor(src8.r/4.f);\n" // dst6r = src8r>>2;
" ocol0.g = frac(src8.r/4.f)*4.f*16.f + floor(src8.g/16.f);\n" // dst6g = ((src8r&0x3)<<4)|(src8g>>4);
" ocol0.b = frac(src8.g/16.f)*16.f*4.f + floor(src8.b/64.f);\n" // dst6b = ((src8g&0xF)<<2)|(src8b>>6);
" ocol0.a = frac(src8.b/64.f)*64.f;\n" // dst6a = src8b&0x3F;
" ocol0 /= 63.f;\n"
"}\n"
"uniform sampler samp0 : register(s0);\n"
"void main(\n"
" out float4 ocol0 : COLOR0,\n"
" in float2 uv0 : TEXCOORD0){\n"
" ocol0 = tex2D(samp0,uv0);\n"
" float4 src8 = round(ocol0*255.f);\n"
" ocol0.r = floor(src8.r/4.f);\n" // dst6r = src8r>>2;
" ocol0.g = frac(src8.r/4.f)*4.f*16.f + floor(src8.g/16.f);\n" // dst6g = ((src8r&0x3)<<4)|(src8g>>4);
" ocol0.b = frac(src8.g/16.f)*16.f*4.f + floor(src8.b/64.f);\n" // dst6b = ((src8g&0xF)<<2)|(src8b>>6);
" ocol0.a = frac(src8.b/64.f)*64.f;\n" // dst6a = src8b&0x3F;
" ocol0 /= 63.f;\n"
"}\n"
};
*/
const char code[] =
{
"uniform sampler samp0 : register(s0);\n"
"void main(\n"
"out float4 ocol0 : COLOR0,\n"
"in float2 uv0 : TEXCOORD0){\n"
"float4 temp1 = float4(1.0f/4.0f,1.0f/16.0f,1.0f/64.0f,0.0f);\n"
"float4 temp2 = float4(1.0f,64.0f,255.0f,1.0f/63.0f);\n"
"float4 src8 = round(tex2D(samp0,uv0)*temp2.z) * temp1;\n"
"ocol0 = (frac(src8.wxyz) * temp2.xyyy + floor(src8)) * temp2.w;\n"
"out float4 ocol0 : COLOR0,\n"
"in float2 uv0 : TEXCOORD0){\n"
"float4 temp1 = float4(1.0f/4.0f,1.0f/16.0f,1.0f/64.0f,0.0f);\n"
"float4 temp2 = float4(1.0f,64.0f,255.0f,1.0f/63.0f);\n"
"float4 src8 = round(tex2D(samp0,uv0)*temp2.z) * temp1;\n"
"ocol0 = (frac(src8.wxyz) * temp2.xyyy + floor(src8)) * temp2.w;\n"
"}\n"
};
if (!s_rgb8_to_rgba6) s_rgb8_to_rgba6 = D3D::CompileAndCreatePixelShader(code, (int)strlen(code));
@ -168,26 +168,28 @@ static LPDIRECT3DPIXELSHADER9 CreateCopyShader(int copyMatrixType, int depthConv
if(copyMatrixType == COPY_TYPE_MATRIXCOLOR)
WRITE(p, "uniform float4 cColMatrix[7] : register(c%d);\n", C_COLORMATRIX);
WRITE(p, "void main(\n"
"out float4 ocol0 : COLOR0,\n");
"out float4 ocol0 : COLOR0,\n");
switch(SSAAMode % MAX_SSAA_SHADERS)
{
case 0: // 1 Sample
WRITE(p, "in float2 uv0 : TEXCOORD0,\n"
"in float uv1 : TEXCOORD1){\n"
"float4 texcol = tex2D(samp0,uv0.xy);\n");
"in float uv1 : TEXCOORD1){\n"
"float4 texcol = tex2D(samp0,uv0.xy);\n");
break;
case 1: // 1 Samples SSAA
WRITE(p, "in float2 uv0 : TEXCOORD0,\n"
"in float uv1 : TEXCOORD1){\n"
"float4 texcol = tex2D(samp0,uv0.xy);\n");
break;
case 2: // 4 Samples SSAA
case 1: // 4 Samples in 4x SSAA buffer
WRITE(p, "in float4 uv0 : TEXCOORD0,\n"
"in float uv1 : TEXCOORD1,\n"
"in float4 uv2 : TEXCOORD2,\n"
"in float4 uv3 : TEXCOORD3){\n"
"float4 texcol = (tex2D(samp0,uv2.xy) + tex2D(samp0,uv2.wz) + tex2D(samp0,uv3.xy) + tex2D(samp0,uv3.wz))*0.25f;\n");
"in float uv1 : TEXCOORD1,\n"
"in float4 uv2 : TEXCOORD2,\n"
"in float4 uv3 : TEXCOORD3){\n"
"float4 texcol = (tex2D(samp0,uv2.xy) + tex2D(samp0,uv2.wz) + tex2D(samp0,uv3.xy) + tex2D(samp0,uv3.wz))*0.25f;\n");
break;
case 2: // 4 Samples in 9x SSAA buffer
WRITE(p, "in float4 uv0 : TEXCOORD0,\n"
"in float uv1 : TEXCOORD1,\n"
"in float4 uv2 : TEXCOORD2,\n"
"in float4 uv3 : TEXCOORD3){\n"
"float4 texcol = (tex2D(samp0,uv2.xy) + tex2D(samp0,uv2.wz) + tex2D(samp0,uv3.xy) + tex2D(samp0,uv3.wz))*0.25f;\n");
break;
}
@ -196,7 +198,7 @@ static LPDIRECT3DPIXELSHADER9 CreateCopyShader(int copyMatrixType, int depthConv
// Watch out for the fire fumes effect in Metroid it's really sensitive to this,
// the lighting in RE0 is also way beyond sensitive since the "good value" is hardcoded and Dolphin is almost always off.
WRITE(p, "float4 EncodedDepth = frac(texcol.r * (16777215.f/16777216.f) * float4(1.0f,256.0f,256.0f*256.0f,1.0f));\n"
"texcol = floor(EncodedDepth * float4(256.f,256.f,256.f,15.0f)) / float4(255.0f,255.0f,255.0f,15.0f);\n");
"texcol = floor(EncodedDepth * float4(256.f,256.f,256.f,15.0f)) / float4(255.0f,255.0f,255.0f,15.0f);\n");
}
else
{
@ -217,7 +219,7 @@ static LPDIRECT3DPIXELSHADER9 CreateCopyShader(int copyMatrixType, int depthConv
WRITE(p, "}\n");
if (text[sizeof(text) - 1] != 0x7C)
PanicAlert("PixelShaderCache copy shader generator - buffer too small, canary has been eaten!");
uselocale(old_locale); // restore locale
freelocale(locale);
return D3D::CompileAndCreatePixelShader(text, (int)strlen(text));
@ -231,10 +233,10 @@ void PixelShaderCache::Init()
{
char pprog[3072];
sprintf(pprog, "void main(\n"
"out float4 ocol0 : COLOR0,\n"
" in float4 incol0 : COLOR0){\n"
"ocol0 = incol0;\n"
"}\n");
"out float4 ocol0 : COLOR0,\n"
" in float4 incol0 : COLOR0){\n"
"ocol0 = incol0;\n"
"}\n");
s_ClearProgram = D3D::CompileAndCreatePixelShader(pprog, (int)strlen(pprog));
}
@ -299,27 +301,27 @@ void PixelShaderCache::Shutdown()
for(int depthType = 0; depthType < NUM_DEPTH_CONVERSION_TYPES; depthType++)
for(int ssaaMode = 0; ssaaMode < MAX_SSAA_SHADERS; ssaaMode++)
if(s_CopyProgram[copyMatrixType][depthType][ssaaMode]
&& (copyMatrixType == 0 || s_CopyProgram[copyMatrixType][depthType][ssaaMode] != s_CopyProgram[copyMatrixType-1][depthType][ssaaMode]))
&& (copyMatrixType == 0 || s_CopyProgram[copyMatrixType][depthType][ssaaMode] != s_CopyProgram[copyMatrixType-1][depthType][ssaaMode]))
s_CopyProgram[copyMatrixType][depthType][ssaaMode]->Release();
for(int copyMatrixType = 0; copyMatrixType < NUM_COPY_TYPES; copyMatrixType++)
for(int depthType = 0; depthType < NUM_DEPTH_CONVERSION_TYPES; depthType++)
for(int ssaaMode = 0; ssaaMode < MAX_SSAA_SHADERS; ssaaMode++)
s_CopyProgram[copyMatrixType][depthType][ssaaMode] = NULL;
for(int copyMatrixType = 0; copyMatrixType < NUM_COPY_TYPES; copyMatrixType++)
for(int depthType = 0; depthType < NUM_DEPTH_CONVERSION_TYPES; depthType++)
for(int ssaaMode = 0; ssaaMode < MAX_SSAA_SHADERS; ssaaMode++)
s_CopyProgram[copyMatrixType][depthType][ssaaMode] = NULL;
if (s_ClearProgram) s_ClearProgram->Release();
s_ClearProgram = NULL;
if (s_rgb8_to_rgba6) s_rgb8_to_rgba6->Release();
s_rgb8_to_rgba6 = NULL;
if (s_rgba6_to_rgb8) s_rgba6_to_rgb8->Release();
s_rgba6_to_rgb8 = NULL;
if (s_ClearProgram) s_ClearProgram->Release();
s_ClearProgram = NULL;
if (s_rgb8_to_rgba6) s_rgb8_to_rgba6->Release();
s_rgb8_to_rgba6 = NULL;
if (s_rgba6_to_rgb8) s_rgba6_to_rgb8->Release();
s_rgba6_to_rgb8 = NULL;
Clear();
g_ps_disk_cache.Sync();
g_ps_disk_cache.Close();
Clear();
g_ps_disk_cache.Sync();
g_ps_disk_cache.Close();
unique_shaders.clear();
unique_shaders.clear();
}
bool PixelShaderCache::SetShader(DSTALPHA_MODE dstAlphaMode, u32 components)

View file

@ -70,7 +70,7 @@ void SetupDeviceObjects()
VertexShaderManager::Dirty();
PixelShaderManager::Dirty();
TextureConverter::Init();
// To avoid shader compilation stutters, read back all shaders from cache.
VertexShaderCache::Init();
PixelShaderCache::Init();
@ -122,7 +122,7 @@ Renderer::Renderer()
fullScreenRes = 0;
D3D::Create(g_ActiveConfig.iAdapter, EmuWindow::GetWnd(),
fullScreenRes, backbuffer_ms_mode, false);
fullScreenRes, backbuffer_ms_mode, false);
IS_AMD = D3D::IsATIDevice();
@ -165,7 +165,7 @@ Renderer::Renderer()
vp.MaxZ = 1.0f;
D3D::dev->SetViewport(&vp);
D3D::dev->Clear(0, NULL, D3DCLEAR_TARGET, 0x0, 0, 0);
D3D::dev->SetRenderTarget(0, FramebufferManager::GetEFBColorRTSurface());
D3D::dev->SetDepthStencilSurface(FramebufferManager::GetEFBDepthRTSurface());
vp.X = 0;
@ -189,7 +189,7 @@ Renderer::~Renderer()
D3D::EndFrame();
D3D::Present();
D3D::Close();
delete[] st;
}
@ -267,7 +267,7 @@ bool Renderer::CheckForResize()
return true;
}
return false;
}
@ -400,7 +400,7 @@ u32 Renderer::AccessEFB(EFBAccessType type, u32 x, u32 y, u32 poke_data)
colmat[0] = colmat[5] = colmat[10] = 1.0f;
PixelShaderManager::SetColorMatrix(colmat); // set transformation
LPDIRECT3DTEXTURE9 read_texture = FramebufferManager::GetEFBDepthTexture();
D3D::ChangeSamplerState(0, D3DSAMP_MINFILTER, D3DTEXF_POINT);
D3DFORMAT bformat = FramebufferManager::GetEFBDepthRTSurfaceFormat();
@ -484,10 +484,10 @@ u32 Renderer::AccessEFB(EFBAccessType type, u32 x, u32 y, u32 poke_data)
// TODO: Speed this up by batching pokes?
ResetAPIState();
D3D::drawColorQuad(poke_data,
(float)RectToLock.left * 2.f / (float)Renderer::GetTargetWidth() - 1.f,
- (float)RectToLock.top * 2.f / (float)Renderer::GetTargetHeight() + 1.f,
(float)RectToLock.right * 2.f / (float)Renderer::GetTargetWidth() - 1.f,
- (float)RectToLock.bottom * 2.f / (float)Renderer::GetTargetHeight() + 1.f);
(float)RectToLock.left * 2.f / (float)Renderer::GetTargetWidth() - 1.f,
- (float)RectToLock.top * 2.f / (float)Renderer::GetTargetHeight() + 1.f,
(float)RectToLock.right * 2.f / (float)Renderer::GetTargetWidth() - 1.f,
- (float)RectToLock.bottom * 2.f / (float)Renderer::GetTargetHeight() + 1.f);
RestoreAPIState();
return 0;
}
@ -797,7 +797,7 @@ void Renderer::Swap(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight,cons
// Prepare to copy the XFBs to our backbuffer
D3D::dev->SetDepthStencilSurface(NULL);
D3D::dev->SetRenderTarget(0, D3D::GetBackBufferSurface());
UpdateDrawRectangle(s_backbuffer_width, s_backbuffer_height);
D3DVIEWPORT9 vp;
@ -856,7 +856,7 @@ void Renderer::Swap(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight,cons
xfbSource = xfbSourceList[i];
MathUtil::Rectangle<float> sourceRc;
sourceRc.left = 0;
sourceRc.top = 0;
sourceRc.right = (float)xfbSource->texWidth;
@ -905,7 +905,7 @@ void Renderer::Swap(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight,cons
Width,Height,
PixelShaderCache::GetColorCopyProgram(g_ActiveConfig.iMultisampleMode),
VertexShaderCache::GetSimpleVertexShader(g_ActiveConfig.iMultisampleMode),Gamma);
}
D3D::RefreshSamplerState(0, D3DSAMP_MINFILTER);
D3D::RefreshSamplerState(0, D3DSAMP_MAGFILTER);
@ -953,7 +953,7 @@ void Renderer::Swap(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight,cons
{
char msg [255];
sprintf_s(msg,255, "Dumping Frames to \"%sframedump0.avi\" (%dx%d RGB24)",
File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), s_recordWidth, s_recordHeight);
File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), s_recordWidth, s_recordHeight);
OSD::AddMessage(msg, 2000);
}
}
@ -1059,7 +1059,7 @@ void Renderer::Swap(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight,cons
s_LastAA = newAA;
UpdateDrawRectangle(s_backbuffer_width, s_backbuffer_height);
int SupersampleCoeficient = (s_LastAA % 3) + 1;
s_LastEFBScale = g_ActiveConfig.iEFBScale;
@ -1081,6 +1081,7 @@ void Renderer::Swap(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight,cons
D3D::dev->SetRenderTarget(0, FramebufferManager::GetEFBColorRTSurface());
D3D::dev->SetDepthStencilSurface(FramebufferManager::GetEFBDepthRTSurface());
D3D::dev->Clear(0, NULL, D3DCLEAR_TARGET | D3DCLEAR_ZBUFFER, D3DCOLOR_XRGB(0,0,0), 1.0f, 0);
SetLineWidth();
}
if (XFBWrited)
@ -1141,8 +1142,8 @@ void Renderer::RestoreState()
D3D::RefreshRenderState(D3DRS_ZFUNC);
}
// TODO: Enable this code. Caused glitches for me however (neobrain)
// for (unsigned int i = 0; i < 8; ++i)
// D3D::dev->SetTexture(i, NULL);
// for (unsigned int i = 0; i < 8; ++i)
// D3D::dev->SetTexture(i, NULL);
}
// ALWAYS call RestoreAPIState for each ResetAPIState call you're doing
@ -1316,9 +1317,8 @@ void Renderer::SetLineWidth()
{
// We can't change line width in D3D unless we use ID3DXLine
float fratio = xfregs.viewport.wd != 0 ? Renderer::EFBToScaledXf(1.f) : 1.0f;
float psize = bpmem.lineptwidth.linesize * fratio / 6.0f;
//little hack to compensate scaling problems in dx9 must be taken out when scaling is fixed.
psize *= 2.0f;
float psize = bpmem.lineptwidth.pointsize * fratio / 6.0f;
psize = psize > 0 ? psize : 1.0;
if (psize > m_fMaxPointSize)
{
psize = m_fMaxPointSize;
@ -1348,7 +1348,7 @@ void Renderer::SetSamplerState(int stage, int texindex)
const FourTexUnits &tex = bpmem.tex[texindex];
const TexMode0 &tm0 = tex.texMode0[stage];
const TexMode1 &tm1 = tex.texMode1[stage];
D3DTEXTUREFILTERTYPE min, mag, mip;
if (g_ActiveConfig.bForceFiltering)
{
@ -1370,7 +1370,7 @@ void Renderer::SetSamplerState(int stage, int texindex)
D3D::SetSamplerState(stage, D3DSAMP_MINFILTER, min);
D3D::SetSamplerState(stage, D3DSAMP_MAGFILTER, mag);
D3D::SetSamplerState(stage, D3DSAMP_MIPFILTER, mip);
D3D::SetSamplerState(stage, D3DSAMP_ADDRESSU, d3dClamps[tm0.wrap_s]);
D3D::SetSamplerState(stage, D3DSAMP_ADDRESSV, d3dClamps[tm0.wrap_t]);

View file

@ -37,15 +37,15 @@ inline void DumpBadShaders()
{
#if defined(_DEBUG) || defined(DEBUGFAST)
// TODO: Reimplement!
/* std::string error_shaders;
/* std::string error_shaders;
error_shaders.append(VertexShaderCache::GetCurrentShaderCode());
error_shaders.append(PixelShaderCache::GetCurrentShaderCode());
char filename[512] = "bad_shader_combo_0.txt";
int which = 0;
while (File::Exists(filename))
{
which++;
sprintf(filename, "bad_shader_combo_%i.txt", which);
which++;
sprintf(filename, "bad_shader_combo_%i.txt", which);
}
File::WriteStringToFile(true, error_shaders, filename);
PanicAlert("DrawIndexedPrimitiveUP failed. Shaders written to %s", filename);*/
@ -64,7 +64,7 @@ void VertexManager::CreateDeviceObjects()
m_index_buffer_size = (IBUFFER_SIZE > DeviceCaps.MaxVertexIndex) ? DeviceCaps.MaxVertexIndex : IBUFFER_SIZE;
//if device caps are not enough for Vbuffer fall back to vertex arrays
if (m_index_buffer_size < MAXIBUFFERSIZE || m_vertex_buffer_size < MAXVBUFFERSIZE) return;
m_vertex_buffers = new LPDIRECT3DVERTEXBUFFER9[MAX_VBUFFER_COUNT];
m_index_buffers = new LPDIRECT3DINDEXBUFFER9[MAX_VBUFFER_COUNT];
@ -144,59 +144,61 @@ void VertexManager::PrepareDrawBuffers(u32 stride)
int datasize = IndexGenerator::GetNumVerts() * stride;
int TdataSize = IndexGenerator::GetTriangleindexLen();
int LDataSize = IndexGenerator::GetLineindexLen();
int PDataSize = IndexGenerator::GetPointindexLen();
int IndexDataSize = TdataSize + LDataSize;
DWORD LockMode = D3DLOCK_NOOVERWRITE;
m_vertex_buffer_cursor--;
m_vertex_buffer_cursor = m_vertex_buffer_cursor - (m_vertex_buffer_cursor % stride) + stride;
if (m_vertex_buffer_cursor > m_vertex_buffer_size - datasize)
if(IndexDataSize)
{
LockMode = D3DLOCK_DISCARD;
m_vertex_buffer_cursor = 0;
m_current_vertex_buffer = (m_current_vertex_buffer + 1) % m_buffers_count;
}
if(FAILED(m_vertex_buffers[m_current_vertex_buffer]->Lock(m_vertex_buffer_cursor, datasize,(VOID**)(&pVertices), LockMode)))
{
DestroyDeviceObjects();
return;
}
memcpy(pVertices, s_pBaseBufferPointer, datasize);
m_vertex_buffers[m_current_vertex_buffer]->Unlock();
DWORD LockMode = D3DLOCK_NOOVERWRITE;
m_vertex_buffer_cursor--;
m_vertex_buffer_cursor = m_vertex_buffer_cursor - (m_vertex_buffer_cursor % stride) + stride;
if (m_vertex_buffer_cursor > m_vertex_buffer_size - datasize)
{
LockMode = D3DLOCK_DISCARD;
m_vertex_buffer_cursor = 0;
m_current_vertex_buffer = (m_current_vertex_buffer + 1) % m_buffers_count;
}
if(FAILED(m_vertex_buffers[m_current_vertex_buffer]->Lock(m_vertex_buffer_cursor, datasize,(VOID**)(&pVertices), LockMode)))
{
DestroyDeviceObjects();
return;
}
memcpy(pVertices, s_pBaseBufferPointer, datasize);
m_vertex_buffers[m_current_vertex_buffer]->Unlock();
LockMode = D3DLOCK_NOOVERWRITE;
if (m_index_buffer_cursor > m_index_buffer_size - IndexDataSize)
{
LockMode = D3DLOCK_DISCARD;
m_index_buffer_cursor = 0;
m_current_index_buffer = (m_current_index_buffer + 1) % m_buffers_count;
}
if(FAILED(m_index_buffers[m_current_index_buffer]->Lock(m_index_buffer_cursor * sizeof(u16), IndexDataSize * sizeof(u16), (VOID**)(&pIndices), LockMode )))
{
DestroyDeviceObjects();
return;
LockMode = D3DLOCK_NOOVERWRITE;
if (m_index_buffer_cursor > m_index_buffer_size - IndexDataSize)
{
LockMode = D3DLOCK_DISCARD;
m_index_buffer_cursor = 0;
m_current_index_buffer = (m_current_index_buffer + 1) % m_buffers_count;
}
if(FAILED(m_index_buffers[m_current_index_buffer]->Lock(m_index_buffer_cursor * sizeof(u16), IndexDataSize * sizeof(u16), (VOID**)(&pIndices), LockMode )))
{
DestroyDeviceObjects();
return;
}
if(TdataSize)
{
memcpy(pIndices, GetTriangleIndexBuffer(), TdataSize * sizeof(u16));
pIndices += TdataSize;
}
if(LDataSize)
{
memcpy(pIndices, GetLineIndexBuffer(), LDataSize * sizeof(u16));
pIndices += LDataSize;
}
m_index_buffers[m_current_index_buffer]->Unlock();
}
if(TdataSize)
{
memcpy(pIndices, GetTriangleIndexBuffer(), TdataSize * sizeof(u16));
pIndices += TdataSize;
}
if(LDataSize)
{
memcpy(pIndices, GetLineIndexBuffer(), LDataSize * sizeof(u16));
pIndices += LDataSize;
}
m_index_buffers[m_current_index_buffer]->Unlock();
if(m_current_stride != stride || m_vertex_buffer_cursor == 0)
{
m_current_stride = stride;
D3D::SetStreamSource( 0, m_vertex_buffers[m_current_vertex_buffer], 0, stride);
D3D::SetStreamSource( 0, m_vertex_buffers[m_current_vertex_buffer], 0, m_current_stride);
}
if (m_index_buffer_cursor == 0)
{
D3D::SetIndices(m_index_buffers[m_current_index_buffer]);
}
ADDSTAT(stats.thisFrame.bytesVertexStreamed, datasize);
ADDSTAT(stats.thisFrame.bytesIndexStreamed, IndexDataSize);
}
@ -241,22 +243,28 @@ void VertexManager::DrawVertexBuffer(int stride)
}
if (points > 0)
{
//DrawIndexedPrimitive does not support point list so we have to draw the points one by one
for (int i = 0; i < points; i++)
//DrawIndexedPrimitive does not support point list so we have to draw them using DrawPrimitive
u16* PointIndexBuffer = GetPointIndexBuffer();
int i = 0;
do
{
int count = i + 1;
while (count < points && PointIndexBuffer[count - 1] + 1 == PointIndexBuffer[count])
{
count++;
}
if (FAILED(D3D::dev->DrawPrimitive(
D3DPT_POINTLIST,
basevertex + GetPointIndexBuffer()[i],
1)))
D3DPT_POINTLIST,
basevertex + PointIndexBuffer[i],
count - i)))
{
DumpBadShaders();
}
INCSTAT(stats.thisFrame.numDrawCalls);
}
i = count;
} while (i < points);
}
}
void VertexManager::DrawVertexArray(int stride)
@ -351,7 +359,7 @@ void VertexManager::vFlush()
PixelShaderManager::SetConstants(g_nativeVertexFmt->m_components);
u32 stride = g_nativeVertexFmt->GetVertexStride();
bool useDstAlpha = !g_ActiveConfig.bDstAlphaPass && bpmem.dstalpha.enable && bpmem.blendmode.alphaupdate &&
bpmem.zcontrol.pixel_format == PIXELFMT_RGBA6_Z24;
bpmem.zcontrol.pixel_format == PIXELFMT_RGBA6_Z24;
bool useDualSource = useDstAlpha && g_ActiveConfig.backend_info.bSupportsDualSourceBlend;
DSTALPHA_MODE AlphaMode = useDualSource ? DSTALPHA_DUAL_SOURCE_BLEND : DSTALPHA_NONE;

View file

@ -59,72 +59,76 @@ void VertexShaderCache::Init()
{
char* vProg = new char[2048];
sprintf(vProg,"struct VSOUTPUT\n"
"{\n"
"float4 vPosition : POSITION;\n"
"float2 vTexCoord : TEXCOORD0;\n"
"float vTexCoord1 : TEXCOORD1;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inTEX1 : TEXCOORD1,float inTEX2 : TEXCOORD2)\n"
"{\n"
"VSOUTPUT OUT;\n"
"OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0;\n"
"OUT.vTexCoord1 = inTEX2;\n"
"return OUT;\n"
"}\n");
"{\n"
"float4 vPosition : POSITION;\n"
"float2 vTexCoord : TEXCOORD0;\n"
"float vTexCoord1 : TEXCOORD1;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inTEX1 : TEXCOORD1,float inTEX2 : TEXCOORD2)\n"
"{\n"
"VSOUTPUT OUT;\n"
"OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0;\n"
"OUT.vTexCoord1 = inTEX2;\n"
"return OUT;\n"
"}\n");
SimpleVertexShader[0] = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg));
sprintf(vProg,"struct VSOUTPUT\n"
"{\n"
"float4 vPosition : POSITION;\n"
"float4 vColor0 : COLOR0;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float4 inColor0: COLOR0)\n"
"{\n"
"VSOUTPUT OUT;\n"
"OUT.vPosition = inPosition;\n"
"OUT.vColor0 = inColor0;\n"
"return OUT;\n"
"}\n");
"{\n"
"float4 vPosition : POSITION;\n"
"float4 vColor0 : COLOR0;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float4 inColor0: COLOR0)\n"
"{\n"
"VSOUTPUT OUT;\n"
"OUT.vPosition = inPosition;\n"
"OUT.vColor0 = inColor0;\n"
"return OUT;\n"
"}\n");
ClearVertexShader = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg));
sprintf(vProg, "struct VSOUTPUT\n"
"{\n"
"float4 vPosition : POSITION;\n"
"float2 vTexCoord : TEXCOORD0;\n"
"float vTexCoord1 : TEXCOORD1;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inInvTexSize : TEXCOORD1,float inTEX2 : TEXCOORD2)\n"
"{\n"
"VSOUTPUT OUT;"
"OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0;\n"
"OUT.vTexCoord1 = inTEX2;\n"
"return OUT;\n"
"}\n");
"{\n"
"float4 vPosition : POSITION;\n"
"float4 vTexCoord : TEXCOORD0;\n"
"float vTexCoord1 : TEXCOORD1;\n"
"float4 vTexCoord2 : TEXCOORD2;\n"
"float4 vTexCoord3 : TEXCOORD3;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inTEX1 : TEXCOORD1,float inTEX2 : TEXCOORD2)\n"
"{\n"
"VSOUTPUT OUT;"
"OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0.xyyx;\n"
"OUT.vTexCoord1 = inTEX2.x;\n"
"OUT.vTexCoord2 = inTEX0.xyyx + (float4(-0.495f,-0.495f, 0.495f,-0.495f) * inTEX1.xyyx);\n"
"OUT.vTexCoord3 = inTEX0.xyyx + (float4( 0.495f, 0.495f,-0.495f, 0.495f) * inTEX1.xyyx);\n"
"return OUT;\n"
"}\n");
SimpleVertexShader[1] = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg));
sprintf(vProg, "struct VSOUTPUT\n"
"{\n"
"float4 vPosition : POSITION;\n"
"float4 vTexCoord : TEXCOORD0;\n"
"float vTexCoord1 : TEXCOORD1;\n"
"float4 vTexCoord2 : TEXCOORD2;\n"
"float4 vTexCoord3 : TEXCOORD3;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inTEX1 : TEXCOORD1,float inTEX2 : TEXCOORD2)\n"
"{\n"
"VSOUTPUT OUT;"
"OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0.xyyx;\n"
"OUT.vTexCoord1 = inTEX2.x;\n"
"OUT.vTexCoord2 = inTEX0.xyyx + (float4(-1.0f,-0.5f, 1.0f,-0.5f) * inTEX1.xyyx);\n"
"OUT.vTexCoord3 = inTEX0.xyyx + (float4( 1.0f, 0.5f,-1.0f, 0.5f) * inTEX1.xyyx);\n"
"return OUT;\n"
"}\n");
"{\n"
"float4 vPosition : POSITION;\n"
"float4 vTexCoord : TEXCOORD0;\n"
"float vTexCoord1 : TEXCOORD1;\n"
"float4 vTexCoord2 : TEXCOORD2;\n"
"float4 vTexCoord3 : TEXCOORD3;\n"
"};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inTEX1 : TEXCOORD1,float inTEX2 : TEXCOORD2)\n"
"{\n"
"VSOUTPUT OUT;"
"OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0.xyyx;\n"
"OUT.vTexCoord1 = inTEX2.x;\n"
"OUT.vTexCoord2 = inTEX0.xyyx + (float4(-0.9f,-0.45f, 0.9f,-0.45f) * inTEX1.xyyx);\n"
"OUT.vTexCoord3 = inTEX0.xyyx + (float4( 0.9f, 0.45f,-0.9f, 0.45f) * inTEX1.xyyx);\n"
"return OUT;\n"
"}\n");
SimpleVertexShader[2] = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg));
Clear();
delete [] vProg;
@ -136,7 +140,7 @@ void VertexShaderCache::Init()
char cache_filename[MAX_PATH];
sprintf(cache_filename, "%sdx9-%s-vs.cache", File::GetUserPath(D_SHADERCACHE_IDX).c_str(),
SConfig::GetInstance().m_LocalCoreStartupParameter.m_strUniqueID.c_str());
SConfig::GetInstance().m_LocalCoreStartupParameter.m_strUniqueID.c_str());
VertexShaderCacheInserter inserter;
g_vs_disk_cache.OpenAndRead(cache_filename, inserter);
@ -168,7 +172,7 @@ void VertexShaderCache::Shutdown()
if (ClearVertexShader)
ClearVertexShader->Release();
ClearVertexShader = NULL;
Clear();
g_vs_disk_cache.Sync();
g_vs_disk_cache.Close();
@ -251,10 +255,16 @@ bool VertexShaderCache::InsertByteCode(const VertexShaderUid &uid, const u8 *byt
return false;
}
float VSConstantbuffer[4*C_VENVCONST_END];
void Renderer::SetVSConstant4f(unsigned int const_number, float f1, float f2, float f3, float f4)
{
const float f[4] = { f1, f2, f3, f4 };
DX9::D3D::dev->SetVertexShaderConstantF(const_number, f, 1);
float* VSConstantbuffer_pointer = &VSConstantbuffer[const_number];
VSConstantbuffer_pointer[0] = f1;
VSConstantbuffer_pointer[1] = f2;
VSConstantbuffer_pointer[2] = f3;
VSConstantbuffer_pointer[3] = f4;
DX9::D3D::dev->SetVertexShaderConstantF(const_number, VSConstantbuffer_pointer, 1);
}
void Renderer::SetVSConstant4fv(unsigned int const_number, const float *f)
@ -264,15 +274,15 @@ void Renderer::SetVSConstant4fv(unsigned int const_number, const float *f)
void Renderer::SetMultiVSConstant3fv(unsigned int const_number, unsigned int count, const float *f)
{
float buf[4*C_VENVCONST_END];
float* VSConstantbuffer_pointer = &VSConstantbuffer[const_number];
for (unsigned int i = 0; i < count; i++)
{
buf[4*i ] = *f++;
buf[4*i+1] = *f++;
buf[4*i+2] = *f++;
buf[4*i+3] = 0.f;
*VSConstantbuffer_pointer++ = *f++;
*VSConstantbuffer_pointer++ = *f++;
*VSConstantbuffer_pointer++ = *f++;
*VSConstantbuffer_pointer++ = 0.f;
}
DX9::D3D::dev->SetVertexShaderConstantF(const_number, buf, count);
DX9::D3D::dev->SetVertexShaderConstantF(const_number, &VSConstantbuffer[const_number], count);
}
void Renderer::SetMultiVSConstant4fv(unsigned int const_number, unsigned int count, const float *f)

View file

@ -90,10 +90,24 @@ void InitBackendInfo()
g_Config.backend_info.bUseRGBATextures = false;
g_Config.backend_info.bUseMinimalMipCount = true;
g_Config.backend_info.bSupports3DVision = true;
g_Config.backend_info.bSupportsPrimitiveRestart = false; // TODO: figure out if it does
g_Config.backend_info.bSupportsPrimitiveRestart = false; // D3D9 does not support primitive restart
g_Config.backend_info.bSupportsSeparateAlphaFunction = device_caps.PrimitiveMiscCaps & D3DPMISCCAPS_SEPARATEALPHABLEND;
// Dual source blend disabled by default until a proper method to test for support is found
g_Config.backend_info.bSupportsDualSourceBlend = false;
g_Config.backend_info.bSupports3DVision = true;
OSVERSIONINFO info;
ZeroMemory(&info, sizeof(OSVERSIONINFO));
info.dwOSVersionInfoSize = sizeof(OSVERSIONINFO);
if (GetVersionEx(&info))
{
// dual source blending is only supported in windows 7 o newer. sorry xp users
// we cannot test for device caps because most drivers just declare the minimun caps
// and don't expose their support for some functionalities
g_Config.backend_info.bSupportsDualSourceBlend = g_Config.backend_info.bSupportsSeparateAlphaFunction && (info.dwPlatformId == VER_PLATFORM_WIN32_NT) && ((info.dwMajorVersion > 6) || ((info.dwMajorVersion == 6) && info.dwMinorVersion >= 1));
}
else
{
g_Config.backend_info.bSupportsDualSourceBlend = false;
}
g_Config.backend_info.bSupportsFormatReinterpretation = true;
g_Config.backend_info.bSupportsPixelLighting = C_PLIGHTS + 40 <= maxConstants && C_PMATERIALS + 4 <= maxConstants;
g_Config.backend_info.bSupportsEarlyZ = false;
@ -112,7 +126,7 @@ void InitBackendInfo()
for (int i = 0; i < (int)adapter.aa_levels.size(); ++i)
g_Config.backend_info.AAModes.push_back(adapter.aa_levels[i].name);
}
// Clear ppshaders string vector
g_Config.backend_info.PPShaders.clear();
@ -139,6 +153,9 @@ bool VideoBackend::Initialize(void *&window_handle)
g_Config.GameIniLoad(SConfig::GetInstance().m_LocalCoreStartupParameter.m_strGameIni.c_str());
g_Config.UpdateProjectionHack();
g_Config.VerifyValidity();
// as only some driver/hardware configurations support dual source blending only enable it if is
// configured by user
g_Config.backend_info.bSupportsDualSourceBlend &= g_Config.bForceDualSourceBlend;
UpdateActiveConfig();
window_handle = (void*)EmuWindow::Create((HWND)window_handle, GetModuleHandle(0), _T("Loading - Please wait."));