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", "EFBScaledCopy", &bCopyEFBScaled, true);
iniFile.Get("Hacks", "EFBCopyCacheEnable", &bEFBCopyCacheEnable, false); iniFile.Get("Hacks", "EFBCopyCacheEnable", &bEFBCopyCacheEnable, false);
iniFile.Get("Hacks", "EFBEmulateFormatChanges", &bEFBEmulateFormatChanges, false); iniFile.Get("Hacks", "EFBEmulateFormatChanges", &bEFBEmulateFormatChanges, false);
iniFile.Get("Hacks", "ForceDualSourceBlend", &bForceDualSourceBlend, false);
iniFile.Get("Hardware", "Adapter", &iAdapter, 0); 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", "EFBScaledCopy", bCopyEFBScaled);
iniFile.Set("Hacks", "EFBCopyCacheEnable", bEFBCopyCacheEnable); iniFile.Set("Hacks", "EFBCopyCacheEnable", bEFBCopyCacheEnable);
iniFile.Set("Hacks", "EFBEmulateFormatChanges", bEFBEmulateFormatChanges); iniFile.Set("Hacks", "EFBEmulateFormatChanges", bEFBEmulateFormatChanges);
iniFile.Set("Hacks", "ForceDualSourceBlend", bForceDualSourceBlend);
iniFile.Set("Hardware", "Adapter", iAdapter); iniFile.Set("Hardware", "Adapter", iAdapter);

View File

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

View File

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

View File

@ -177,12 +177,14 @@ static LPDIRECT3DPIXELSHADER9 CreateCopyShader(int copyMatrixType, int depthConv
"in float uv1 : TEXCOORD1){\n" "in float uv1 : TEXCOORD1){\n"
"float4 texcol = tex2D(samp0,uv0.xy);\n"); "float4 texcol = tex2D(samp0,uv0.xy);\n");
break; break;
case 1: // 1 Samples SSAA case 1: // 4 Samples in 4x SSAA buffer
WRITE(p, "in float2 uv0 : TEXCOORD0,\n" WRITE(p, "in float4 uv0 : TEXCOORD0,\n"
"in float uv1 : TEXCOORD1){\n" "in float uv1 : TEXCOORD1,\n"
"float4 texcol = tex2D(samp0,uv0.xy);\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; break;
case 2: // 4 Samples SSAA case 2: // 4 Samples in 9x SSAA buffer
WRITE(p, "in float4 uv0 : TEXCOORD0,\n" WRITE(p, "in float4 uv0 : TEXCOORD0,\n"
"in float uv1 : TEXCOORD1,\n" "in float uv1 : TEXCOORD1,\n"
"in float4 uv2 : TEXCOORD2,\n" "in float4 uv2 : TEXCOORD2,\n"

View File

@ -1081,6 +1081,7 @@ void Renderer::Swap(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight,cons
D3D::dev->SetRenderTarget(0, FramebufferManager::GetEFBColorRTSurface()); D3D::dev->SetRenderTarget(0, FramebufferManager::GetEFBColorRTSurface());
D3D::dev->SetDepthStencilSurface(FramebufferManager::GetEFBDepthRTSurface()); D3D::dev->SetDepthStencilSurface(FramebufferManager::GetEFBDepthRTSurface());
D3D::dev->Clear(0, NULL, D3DCLEAR_TARGET | D3DCLEAR_ZBUFFER, D3DCOLOR_XRGB(0,0,0), 1.0f, 0); D3D::dev->Clear(0, NULL, D3DCLEAR_TARGET | D3DCLEAR_ZBUFFER, D3DCOLOR_XRGB(0,0,0), 1.0f, 0);
SetLineWidth();
} }
if (XFBWrited) if (XFBWrited)
@ -1316,9 +1317,8 @@ void Renderer::SetLineWidth()
{ {
// We can't change line width in D3D unless we use ID3DXLine // 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 fratio = xfregs.viewport.wd != 0 ? Renderer::EFBToScaledXf(1.f) : 1.0f;
float psize = bpmem.lineptwidth.linesize * fratio / 6.0f; float psize = bpmem.lineptwidth.pointsize * fratio / 6.0f;
//little hack to compensate scaling problems in dx9 must be taken out when scaling is fixed. psize = psize > 0 ? psize : 1.0;
psize *= 2.0f;
if (psize > m_fMaxPointSize) if (psize > m_fMaxPointSize)
{ {
psize = m_fMaxPointSize; psize = m_fMaxPointSize;

View File

@ -144,8 +144,9 @@ void VertexManager::PrepareDrawBuffers(u32 stride)
int datasize = IndexGenerator::GetNumVerts() * stride; int datasize = IndexGenerator::GetNumVerts() * stride;
int TdataSize = IndexGenerator::GetTriangleindexLen(); int TdataSize = IndexGenerator::GetTriangleindexLen();
int LDataSize = IndexGenerator::GetLineindexLen(); int LDataSize = IndexGenerator::GetLineindexLen();
int PDataSize = IndexGenerator::GetPointindexLen();
int IndexDataSize = TdataSize + LDataSize; int IndexDataSize = TdataSize + LDataSize;
if(IndexDataSize)
{
DWORD LockMode = D3DLOCK_NOOVERWRITE; DWORD LockMode = D3DLOCK_NOOVERWRITE;
m_vertex_buffer_cursor--; m_vertex_buffer_cursor--;
m_vertex_buffer_cursor = m_vertex_buffer_cursor - (m_vertex_buffer_cursor % stride) + stride; m_vertex_buffer_cursor = m_vertex_buffer_cursor - (m_vertex_buffer_cursor % stride) + stride;
@ -187,10 +188,11 @@ void VertexManager::PrepareDrawBuffers(u32 stride)
pIndices += LDataSize; pIndices += LDataSize;
} }
m_index_buffers[m_current_index_buffer]->Unlock(); m_index_buffers[m_current_index_buffer]->Unlock();
}
if(m_current_stride != stride || m_vertex_buffer_cursor == 0) if(m_current_stride != stride || m_vertex_buffer_cursor == 0)
{ {
m_current_stride = stride; 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) if (m_index_buffer_cursor == 0)
{ {
@ -241,20 +243,26 @@ void VertexManager::DrawVertexBuffer(int stride)
} }
if (points > 0) if (points > 0)
{ {
//DrawIndexedPrimitive does not support point list so we have to draw the points one by one //DrawIndexedPrimitive does not support point list so we have to draw them using DrawPrimitive
for (int i = 0; i < points; i++) 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( if (FAILED(D3D::dev->DrawPrimitive(
D3DPT_POINTLIST, D3DPT_POINTLIST,
basevertex + GetPointIndexBuffer()[i], basevertex + PointIndexBuffer[i],
1))) count - i)))
{ {
DumpBadShaders(); DumpBadShaders();
} }
INCSTAT(stats.thisFrame.numDrawCalls); INCSTAT(stats.thisFrame.numDrawCalls);
} i = count;
} while (i < points);
} }
} }

View File

@ -92,15 +92,19 @@ void VertexShaderCache::Init()
sprintf(vProg, "struct VSOUTPUT\n" sprintf(vProg, "struct VSOUTPUT\n"
"{\n" "{\n"
"float4 vPosition : POSITION;\n" "float4 vPosition : POSITION;\n"
"float2 vTexCoord : TEXCOORD0;\n" "float4 vTexCoord : TEXCOORD0;\n"
"float vTexCoord1 : TEXCOORD1;\n" "float vTexCoord1 : TEXCOORD1;\n"
"float4 vTexCoord2 : TEXCOORD2;\n"
"float4 vTexCoord3 : TEXCOORD3;\n"
"};\n" "};\n"
"VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inInvTexSize : TEXCOORD1,float inTEX2 : TEXCOORD2)\n" "VSOUTPUT main(float4 inPosition : POSITION,float2 inTEX0 : TEXCOORD0,float2 inTEX1 : TEXCOORD1,float inTEX2 : TEXCOORD2)\n"
"{\n" "{\n"
"VSOUTPUT OUT;" "VSOUTPUT OUT;"
"OUT.vPosition = inPosition;\n" "OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0;\n" "OUT.vTexCoord = inTEX0.xyyx;\n"
"OUT.vTexCoord1 = inTEX2;\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" "return OUT;\n"
"}\n"); "}\n");
SimpleVertexShader[1] = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg)); SimpleVertexShader[1] = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg));
@ -119,8 +123,8 @@ void VertexShaderCache::Init()
"OUT.vPosition = inPosition;\n" "OUT.vPosition = inPosition;\n"
"OUT.vTexCoord = inTEX0.xyyx;\n" "OUT.vTexCoord = inTEX0.xyyx;\n"
"OUT.vTexCoord1 = inTEX2.x;\n" "OUT.vTexCoord1 = inTEX2.x;\n"
"OUT.vTexCoord2 = inTEX0.xyyx + (float4(-1.0f,-0.5f, 1.0f,-0.5f) * inTEX1.xyyx);\n" "OUT.vTexCoord2 = inTEX0.xyyx + (float4(-0.9f,-0.45f, 0.9f,-0.45f) * inTEX1.xyyx);\n"
"OUT.vTexCoord3 = inTEX0.xyyx + (float4( 1.0f, 0.5f,-1.0f, 0.5f) * inTEX1.xyyx);\n" "OUT.vTexCoord3 = inTEX0.xyyx + (float4( 0.9f, 0.45f,-0.9f, 0.45f) * inTEX1.xyyx);\n"
"return OUT;\n" "return OUT;\n"
"}\n"); "}\n");
SimpleVertexShader[2] = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg)); SimpleVertexShader[2] = D3D::CompileAndCreateVertexShader(vProg, (int)strlen(vProg));
@ -251,10 +255,16 @@ bool VertexShaderCache::InsertByteCode(const VertexShaderUid &uid, const u8 *byt
return false; return false;
} }
float VSConstantbuffer[4*C_VENVCONST_END];
void Renderer::SetVSConstant4f(unsigned int const_number, float f1, float f2, float f3, float f4) void Renderer::SetVSConstant4f(unsigned int const_number, float f1, float f2, float f3, float f4)
{ {
const float f[4] = { f1, f2, f3, f4 }; float* VSConstantbuffer_pointer = &VSConstantbuffer[const_number];
DX9::D3D::dev->SetVertexShaderConstantF(const_number, f, 1); 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) 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) 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++) for (unsigned int i = 0; i < count; i++)
{ {
buf[4*i ] = *f++; *VSConstantbuffer_pointer++ = *f++;
buf[4*i+1] = *f++; *VSConstantbuffer_pointer++ = *f++;
buf[4*i+2] = *f++; *VSConstantbuffer_pointer++ = *f++;
buf[4*i+3] = 0.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) 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.bUseRGBATextures = false;
g_Config.backend_info.bUseMinimalMipCount = true; g_Config.backend_info.bUseMinimalMipCount = true;
g_Config.backend_info.bSupports3DVision = 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; 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 // Dual source blend disabled by default until a proper method to test for support is found
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.bSupportsDualSourceBlend = false;
}
g_Config.backend_info.bSupportsFormatReinterpretation = true; g_Config.backend_info.bSupportsFormatReinterpretation = true;
g_Config.backend_info.bSupportsPixelLighting = C_PLIGHTS + 40 <= maxConstants && C_PMATERIALS + 4 <= maxConstants; g_Config.backend_info.bSupportsPixelLighting = C_PLIGHTS + 40 <= maxConstants && C_PMATERIALS + 4 <= maxConstants;
g_Config.backend_info.bSupportsEarlyZ = false; g_Config.backend_info.bSupportsEarlyZ = false;
@ -139,6 +153,9 @@ bool VideoBackend::Initialize(void *&window_handle)
g_Config.GameIniLoad(SConfig::GetInstance().m_LocalCoreStartupParameter.m_strGameIni.c_str()); g_Config.GameIniLoad(SConfig::GetInstance().m_LocalCoreStartupParameter.m_strGameIni.c_str());
g_Config.UpdateProjectionHack(); g_Config.UpdateProjectionHack();
g_Config.VerifyValidity(); 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(); UpdateActiveConfig();
window_handle = (void*)EmuWindow::Create((HWND)window_handle, GetModuleHandle(0), _T("Loading - Please wait.")); window_handle = (void*)EmuWindow::Create((HWND)window_handle, GetModuleHandle(0), _T("Loading - Please wait."));