void VertexShaderCache::Init()
{
const D3D11_INPUT_ELEMENT_DESC simpleelems[2] = {
{"POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0},
{"TEXCOORD", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0},
};
const D3D11_INPUT_ELEMENT_DESC clearelems[2] = {
{"POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0},
{"COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0},
};
unsigned int cbsize = ROUND_UP(sizeof(VertexShaderConstants), 16); // must be a multiple of 16
D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(cbsize, D3D11_BIND_CONSTANT_BUFFER,
D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
HRESULT hr = D3D::device->CreateBuffer(&cbdesc, nullptr, &vscbuf);
CHECK(hr == S_OK, "Create vertex shader constant buffer (size=%u)", cbsize);
D3D::SetDebugObjectName((ID3D11DeviceChild*)vscbuf,
"vertex shader constant buffer used to emulate the GX pipeline");
D3DBlob* blob;
D3D::CompileVertexShader(simple_shader_code, &blob);
D3D::device->CreateInputLayout(simpleelems, 2, blob->Data(), blob->Size(), &SimpleLayout);
SimpleVertexShader = D3D::CreateVertexShaderFromByteCode(blob);
if (SimpleLayout == nullptr || SimpleVertexShader == nullptr)
PanicAlert("Failed to create simple vertex shader or input layout at %s %d\n", __FILE__,
__LINE__);
blob->Release();
D3D::SetDebugObjectName((ID3D11DeviceChild*)SimpleVertexShader, "simple vertex shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)SimpleLayout, "simple input layout");
D3D::CompileVertexShader(clear_shader_code, &blob);
D3D::device->CreateInputLayout(clearelems, 2, blob->Data(), blob->Size(), &ClearLayout);
ClearVertexShader = D3D::CreateVertexShaderFromByteCode(blob);
if (ClearLayout == nullptr || ClearVertexShader == nullptr)
PanicAlert("Failed to create clear vertex shader or input layout at %s %d\n", __FILE__,
__LINE__);
blob->Release();
D3D::SetDebugObjectName((ID3D11DeviceChild*)ClearVertexShader, "clear vertex shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)ClearLayout, "clear input layout");
Clear();
if (!File::Exists(File::GetUserPath(D_SHADERCACHE_IDX)))
File::CreateDir(File::GetUserPath(D_SHADERCACHE_IDX));
SETSTAT(stats.numVertexShadersCreated, 0);
SETSTAT(stats.numVertexShadersAlive, 0);
std::string cache_filename =
StringFromFormat("%sdx11-%s-vs.cache", File::GetUserPath(D_SHADERCACHE_IDX).c_str(),
SConfig::GetInstance().m_strUniqueID.c_str());
VertexShaderCacheInserter inserter;
g_vs_disk_cache.OpenAndRead(cache_filename, inserter);
if (g_Config.bEnableShaderDebugging)
Clear();
last_entry = nullptr;
}
void DX11::Shader::UpdateConstantBuffer(ID3D11DeviceContext * context, std::vector<float> & constants, ID3D11Buffer ** buffer)
{
while((constants.size() % 4) != 0)
{
constants.push_back(0.0f);
}
if(*buffer)
{
context->UpdateSubresource(*buffer, 0, 0, constants.data(), 0, 0);
}
else
{
ID3D11Device * device = 0;
context->GetDevice(&device);
if(device)
{
D3D11_SUBRESOURCE_DATA data = { 0 };
data.pSysMem = constants.data();
device->CreateBuffer(
&CD3D11_BUFFER_DESC(sizeof(float) * constants.size(), D3D11_BIND_CONSTANT_BUFFER),
&data,
buffer);
device->Release();
}
}
}
void GeometryShaderCache::Init()
{
unsigned int gbsize = Common::AlignUp(static_cast<unsigned int>(sizeof(GeometryShaderConstants)),
16); // must be a multiple of 16
D3D11_BUFFER_DESC gbdesc = CD3D11_BUFFER_DESC(gbsize, D3D11_BIND_CONSTANT_BUFFER,
D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
HRESULT hr = D3D::device->CreateBuffer(&gbdesc, nullptr, &gscbuf);
CHECK(hr == S_OK, "Create geometry shader constant buffer (size=%u)", gbsize);
D3D::SetDebugObjectName((ID3D11DeviceChild*)gscbuf,
"geometry shader constant buffer used to emulate the GX pipeline");
// used when drawing clear quads
ClearGeometryShader = D3D::CompileAndCreateGeometryShader(clear_shader_code);
CHECK(ClearGeometryShader != nullptr, "Create clear geometry shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)ClearGeometryShader, "clear geometry shader");
// used for buffer copy
CopyGeometryShader = D3D::CompileAndCreateGeometryShader(copy_shader_code);
CHECK(CopyGeometryShader != nullptr, "Create copy geometry shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)CopyGeometryShader, "copy geometry shader");
Clear();
if (g_ActiveConfig.bShaderCache)
LoadShaderCache();
}
void GeometryShaderCache::Init()
{
unsigned int gbsize = ROUND_UP(sizeof(GeometryShaderConstants), 16); // must be a multiple of 16
D3D11_BUFFER_DESC gbdesc = CD3D11_BUFFER_DESC(gbsize, D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
HRESULT hr = D3D::device->CreateBuffer(&gbdesc, nullptr, &gscbuf);
CHECK(hr == S_OK, "Create geometry shader constant buffer (size=%u)", gbsize);
D3D::SetDebugObjectName((ID3D11DeviceChild*)gscbuf, "geometry shader constant buffer used to emulate the GX pipeline");
// used when drawing clear quads
ClearGeometryShader = D3D::CompileAndCreateGeometryShader(clear_shader_code);
CHECK(ClearGeometryShader != nullptr, "Create clear geometry shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)ClearGeometryShader, "clear geometry shader");
// used for buffer copy
CopyGeometryShader = D3D::CompileAndCreateGeometryShader(copy_shader_code);
CHECK(CopyGeometryShader != nullptr, "Create copy geometry shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)CopyGeometryShader, "copy geometry shader");
Clear();
if (!File::Exists(File::GetUserPath(D_SHADERCACHE_IDX)))
File::CreateDir(File::GetUserPath(D_SHADERCACHE_IDX));
std::string cache_filename = StringFromFormat("%sdx11-%s-gs.cache", File::GetUserPath(D_SHADERCACHE_IDX).c_str(),
SConfig::GetInstance().m_strUniqueID.c_str());
GeometryShaderCacheInserter inserter;
g_gs_disk_cache.OpenAndRead(cache_filename, inserter);
if (g_Config.bEnableShaderDebugging)
Clear();
last_entry = nullptr;
}
void TextureCache::TCacheEntry::FromRenderTarget(u8* dst, unsigned int dstFormat, u32 dstStride,
PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool isIntensity, bool scaleByHalf, u32 cbufid,
const float *colmat)
{
g_renderer->ResetAPIState();
// stretch picture with increased internal resolution
const D3D11_VIEWPORT vp = CD3D11_VIEWPORT(0.f, 0.f, (float)config.width, (float)config.height);
D3D::context->RSSetViewports(1, &vp);
// set transformation
if (nullptr == efbcopycbuf[cbufid].get())
{
const D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(28 * sizeof(float), D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DEFAULT);
D3D11_SUBRESOURCE_DATA data;
data.pSysMem = colmat;
HRESULT hr = D3D::device->CreateBuffer(&cbdesc, &data, D3D::ToAddr(efbcopycbuf[cbufid]));
CHECK(SUCCEEDED(hr), "Create efb copy constant buffer %d", cbufid);
D3D::SetDebugObjectName(efbcopycbuf[cbufid].get(), "a constant buffer used in TextureCache::CopyRenderTargetToTexture");
}
D3D::stateman->SetPixelConstants(efbcopycbuf[cbufid].get());
const TargetRectangle targetSource = g_renderer->ConvertEFBRectangle(srcRect);
// TODO: try targetSource.asRECT();
const D3D11_RECT sourcerect = CD3D11_RECT(targetSource.left, targetSource.top, targetSource.right, targetSource.bottom);
// Use linear filtering if (bScaleByHalf), use point filtering otherwise
if (scaleByHalf)
D3D::SetLinearCopySampler();
else
D3D::SetPointCopySampler();
// if texture is currently in use, it needs to be temporarily unset
u32 textureSlotMask = D3D::stateman->UnsetTexture(texture->GetSRV());
D3D::stateman->Apply();
D3D::context->OMSetRenderTargets(1, &texture->GetRTV(), nullptr);
// Create texture copy
D3D::drawShadedTexQuad(
((srcFormat == PEControl::Z24) ? FramebufferManager::GetEFBDepthTexture() : FramebufferManager::GetEFBColorTexture())->GetSRV(),
&sourcerect, Renderer::GetTargetWidth(), Renderer::GetTargetHeight(),
(srcFormat == PEControl::Z24) ? PixelShaderCache::GetDepthMatrixProgram(true) : PixelShaderCache::GetColorMatrixProgram(true),
VertexShaderCache::GetSimpleVertexShader(),
VertexShaderCache::GetSimpleInputLayout(),
(g_Config.iStereoMode > 0) ? GeometryShaderCache::GetCopyGeometryShader() : nullptr);
D3D::context->OMSetRenderTargets(1, &FramebufferManager::GetEFBColorTexture()->GetRTV(), FramebufferManager::GetEFBDepthTexture()->GetDSV());
g_renderer->RestoreAPIState();
if (g_ActiveConfig.bSkipEFBCopyToRam)
{
this->Zero(dst);
}
else
{
s_encoder->Encode(dst, this, srcFormat, srcRect, isIntensity, scaleByHalf);
}
}
void VertexShaderCache::Init()
{
const D3D11_INPUT_ELEMENT_DESC simpleelems[2] = {
{"POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0},
{"TEXCOORD", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0},
};
const D3D11_INPUT_ELEMENT_DESC clearelems[2] = {
{"POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0},
{"COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0},
};
unsigned int cbsize = Common::AlignUp(static_cast<unsigned int>(sizeof(VertexShaderConstants)),
16); // must be a multiple of 16
D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(cbsize, D3D11_BIND_CONSTANT_BUFFER,
D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
HRESULT hr = D3D::device->CreateBuffer(&cbdesc, nullptr, &vscbuf);
CHECK(hr == S_OK, "Create vertex shader constant buffer (size=%u)", cbsize);
D3D::SetDebugObjectName(vscbuf, "vertex shader constant buffer used to emulate the GX pipeline");
D3DBlob* blob;
D3D::CompileVertexShader(simple_shader_code, &blob);
D3D::device->CreateInputLayout(simpleelems, 2, blob->Data(), blob->Size(), &SimpleLayout);
SimpleVertexShader = D3D::CreateVertexShaderFromByteCode(blob);
if (SimpleLayout == nullptr || SimpleVertexShader == nullptr)
PanicAlert("Failed to create simple vertex shader or input layout at %s %d\n", __FILE__,
__LINE__);
blob->Release();
D3D::SetDebugObjectName(SimpleVertexShader, "simple vertex shader");
D3D::SetDebugObjectName(SimpleLayout, "simple input layout");
D3D::CompileVertexShader(clear_shader_code, &blob);
D3D::device->CreateInputLayout(clearelems, 2, blob->Data(), blob->Size(), &ClearLayout);
ClearVertexShader = D3D::CreateVertexShaderFromByteCode(blob);
if (ClearLayout == nullptr || ClearVertexShader == nullptr)
PanicAlert("Failed to create clear vertex shader or input layout at %s %d\n", __FILE__,
__LINE__);
blob->Release();
D3D::SetDebugObjectName(ClearVertexShader, "clear vertex shader");
D3D::SetDebugObjectName(ClearLayout, "clear input layout");
Clear();
SETSTAT(stats.numVertexShadersCreated, 0);
SETSTAT(stats.numVertexShadersAlive, 0);
if (g_ActiveConfig.bShaderCache)
LoadShaderCache();
g_async_compiler = std::make_unique<VideoCommon::AsyncShaderCompiler>();
g_async_compiler->ResizeWorkerThreads(g_ActiveConfig.CanPrecompileUberShaders() ?
g_ActiveConfig.GetShaderPrecompilerThreads() :
g_ActiveConfig.GetShaderCompilerThreads());
if (g_ActiveConfig.CanPrecompileUberShaders())
QueueUberShaderCompiles();
}
void InitUtils()
{
util_vbuf = new UtilVertexBuffer(0x4000);
float border[4] = { 0.f, 0.f, 0.f, 0.f };
D3D11_SAMPLER_DESC samDesc = CD3D11_SAMPLER_DESC(D3D11_FILTER_MIN_MAG_MIP_POINT, D3D11_TEXTURE_ADDRESS_BORDER, D3D11_TEXTURE_ADDRESS_BORDER, D3D11_TEXTURE_ADDRESS_BORDER, 0.f, 1, D3D11_COMPARISON_ALWAYS, border, 0.f, 0.f);
HRESULT hr = D3D::device->CreateSamplerState(&samDesc, &point_copy_sampler);
if (FAILED(hr)) PanicAlert("Failed to create sampler state at %s %d\n", __FILE__, __LINE__);
else SetDebugObjectName((ID3D11DeviceChild*)point_copy_sampler, "point copy sampler state");
samDesc = CD3D11_SAMPLER_DESC(D3D11_FILTER_MIN_MAG_MIP_LINEAR, D3D11_TEXTURE_ADDRESS_BORDER, D3D11_TEXTURE_ADDRESS_BORDER, D3D11_TEXTURE_ADDRESS_BORDER, 0.f, 1, D3D11_COMPARISON_ALWAYS, border, 0.f, 0.f);
hr = D3D::device->CreateSamplerState(&samDesc, &linear_copy_sampler);
if (FAILED(hr)) PanicAlert("Failed to create sampler state at %s %d\n", __FILE__, __LINE__);
else SetDebugObjectName((ID3D11DeviceChild*)linear_copy_sampler, "linear copy sampler state");
// cached data used to avoid unnecessarily reloading the vertex buffers
memset(&tex_quad_data, 0, sizeof(tex_quad_data));
memset(&tex_sub_quad_data, 0, sizeof(tex_sub_quad_data));
memset(&draw_quad_data, 0, sizeof(draw_quad_data));
memset(&clear_quad_data, 0, sizeof(clear_quad_data));
// make sure to properly load the vertex data whenever the corresponding functions get called the first time
stq_observer = stsq_observer = cq_observer = clearq_observer = true;
util_vbuf->AddWrapObserver(&stq_observer);
util_vbuf->AddWrapObserver(&stsq_observer);
util_vbuf->AddWrapObserver(&cq_observer);
util_vbuf->AddWrapObserver(&clearq_observer);
font.Init();
// Create resources for encoder quads
// Create vertex quad
D3D11_BUFFER_DESC bd = CD3D11_BUFFER_DESC(sizeof(ENCODER_QUAD_VERTS),
D3D11_BIND_VERTEX_BUFFER, D3D11_USAGE_IMMUTABLE);
D3D11_SUBRESOURCE_DATA srd = { ENCODER_QUAD_VERTS, 0, 0 };
hr = D3D::device->CreateBuffer(&bd, &srd, &s_encoderQuad);
CHECK(SUCCEEDED(hr), "create encoder quad buffer");
// Create vertex shader
D3DBlob* blob;
D3D::CompileVertexShader(ENCODER_VS, sizeof(ENCODER_VS), &blob);
s_encoderVShader = D3D::CreateVertexShaderFromByteCode(blob);
CHECK(SUCCEEDED(hr), "create encoder vertex shader");
// Create input layout
hr = D3D::device->CreateInputLayout(ENCODER_QUAD_LAYOUT_DESC,
sizeof(ENCODER_QUAD_LAYOUT_DESC) / sizeof(D3D11_INPUT_ELEMENT_DESC),
blob->Data(), blob->Size(), &s_encoderQuadLayout);
CHECK(SUCCEEDED(hr), "create encoder input layout");
blob->Release();
// Create vertex shader for encoder quads with texture coords
s_encoderTexVShader = D3D::CompileAndCreateVertexShader(ENCODER_TEX_VS, sizeof(ENCODER_TEX_VS));
}
ConstantBufferPtr ConstantBuffer::Create(ID3D11Device* pDevice, UINT size)
{
ConstantBufferPtr p(new ConstantBuffer);
D3D11_BUFFER_DESC bd = CD3D11_BUFFER_DESC((size+15)&~15,
D3D11_BIND_CONSTANT_BUFFER);
CHECK_HR(pDevice->CreateBuffer(&bd, NULL, &p->m_pBuffer));
return p;
}
DX11::TextureShader::TextureShader(ID3D11Device * device) :
DX11::Shader(device, Type::Texture), standardConstBuffer(0), pixelObject(0)
{
device->CreateBuffer(
&CD3D11_BUFFER_DESC(sizeof(StandardConstants), D3D11_BIND_CONSTANT_BUFFER),
0,
&standardConstBuffer);
ZeroMemory(paramConstBuffers, NUM_PARAM_BUFFERS * sizeof(void*));
}
DX11::ModelShader::ModelShader(ID3D11Device * device)
: DX11::Shader(device, Type::Model)
, vertexConstBuffer(0)
, pixelConstBuffer(0)
, indirectArgsBuffer(0)
, currentTechnique(0)
, indirectPrimitive(PrimitiveUnknown)
{
device->CreateBuffer(
&CD3D11_BUFFER_DESC(sizeof(VertexConstants), D3D11_BIND_CONSTANT_BUFFER),
0,
&vertexConstBuffer);
device->CreateBuffer(
&CD3D11_BUFFER_DESC(sizeof(PixelConstants), D3D11_BIND_CONSTANT_BUFFER),
0,
&pixelConstBuffer);
device->CreateBuffer(
&CD3D11_BUFFER_DESC(sizeof(LightConstants)* MAX_LIGHTS, D3D11_BIND_CONSTANT_BUFFER),
0,
&lightParamsBuffer);
}
void PixelShaderCache::Init()
{
unsigned int cbsize = Common::AlignUp(static_cast<unsigned int>(sizeof(PixelShaderConstants)),
16); // must be a multiple of 16
D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(cbsize, D3D11_BIND_CONSTANT_BUFFER,
D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
D3D::device->CreateBuffer(&cbdesc, nullptr, &pscbuf);
CHECK(pscbuf != nullptr, "Create pixel shader constant buffer");
D3D::SetDebugObjectName((ID3D11DeviceChild*)pscbuf,
"pixel shader constant buffer used to emulate the GX pipeline");
// used when drawing clear quads
s_ClearProgram = D3D::CompileAndCreatePixelShader(clear_program_code);
CHECK(s_ClearProgram != nullptr, "Create clear pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_ClearProgram, "clear pixel shader");
// used for anaglyph stereoscopy
s_AnaglyphProgram = D3D::CompileAndCreatePixelShader(anaglyph_program_code);
CHECK(s_AnaglyphProgram != nullptr, "Create anaglyph pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_AnaglyphProgram, "anaglyph pixel shader");
// used when copying/resolving the color buffer
s_ColorCopyProgram[0] = D3D::CompileAndCreatePixelShader(color_copy_program_code);
CHECK(s_ColorCopyProgram[0] != nullptr, "Create color copy pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_ColorCopyProgram[0], "color copy pixel shader");
// used for color conversion
s_ColorMatrixProgram[0] = D3D::CompileAndCreatePixelShader(color_matrix_program_code);
CHECK(s_ColorMatrixProgram[0] != nullptr, "Create color matrix pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_ColorMatrixProgram[0], "color matrix pixel shader");
// used for depth copy
s_DepthMatrixProgram[0] = D3D::CompileAndCreatePixelShader(depth_matrix_program);
CHECK(s_DepthMatrixProgram[0] != nullptr, "Create depth matrix pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_DepthMatrixProgram[0], "depth matrix pixel shader");
Clear();
if (!File::Exists(File::GetUserPath(D_SHADERCACHE_IDX)))
File::CreateDir(File::GetUserPath(D_SHADERCACHE_IDX));
SETSTAT(stats.numPixelShadersCreated, 0);
SETSTAT(stats.numPixelShadersAlive, 0);
std::string cache_filename =
StringFromFormat("%sdx11-%s-ps.cache", File::GetUserPath(D_SHADERCACHE_IDX).c_str(),
SConfig::GetInstance().m_strGameID.c_str());
PixelShaderCacheInserter inserter;
g_ps_disk_cache.OpenAndRead(cache_filename, inserter);
last_entry = nullptr;
}
void TextureCache::TCacheEntry::FromRenderTarget(u8* dst, PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool scaleByHalf, unsigned int cbufid, const float *colmat)
{
g_renderer->ResetAPIState();
// stretch picture with increased internal resolution
const D3D11_VIEWPORT vp = CD3D11_VIEWPORT(0.f, 0.f, (float)config.width, (float)config.height);
D3D::context->RSSetViewports(1, &vp);
// set transformation
if (nullptr == efbcopycbuf[cbufid])
{
const D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(28 * sizeof(float), D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DEFAULT);
D3D11_SUBRESOURCE_DATA data;
data.pSysMem = colmat;
HRESULT hr = D3D::device->CreateBuffer(&cbdesc, &data, &efbcopycbuf[cbufid]);
CHECK(SUCCEEDED(hr), "Create efb copy constant buffer %d", cbufid);
D3D::SetDebugObjectName((ID3D11DeviceChild*)efbcopycbuf[cbufid], "a constant buffer used in TextureCache::CopyRenderTargetToTexture");
}
D3D::stateman->SetPixelConstants(efbcopycbuf[cbufid]);
const TargetRectangle targetSource = g_renderer->ConvertEFBRectangle(srcRect);
// TODO: try targetSource.asRECT();
const D3D11_RECT sourcerect = CD3D11_RECT(targetSource.left, targetSource.top, targetSource.right, targetSource.bottom);
// Use linear filtering if (bScaleByHalf), use point filtering otherwise
if (scaleByHalf)
D3D::SetLinearCopySampler();
else
D3D::SetPointCopySampler();
// Make sure we don't draw with the texture set as both a source and target.
// (This can happen because we don't unbind textures when we free them.)
D3D::stateman->UnsetTexture(texture->GetSRV());
D3D::context->OMSetRenderTargets(1, &texture->GetRTV(), nullptr);
// Create texture copy
D3D::drawShadedTexQuad(
(srcFormat == PEControl::Z24 ? FramebufferManager::GetEFBDepthTexture() : FramebufferManager::GetEFBColorTexture())->GetSRV(),
&sourcerect, Renderer::GetTargetWidth(),
Renderer::GetTargetHeight(),
srcFormat == PEControl::Z24 ? PixelShaderCache::GetDepthMatrixProgram(true) : PixelShaderCache::GetColorMatrixProgram(true),
VertexShaderCache::GetSimpleVertexShader(),
VertexShaderCache::GetSimpleInputLayout(),
GeometryShaderCache::GetCopyGeometryShader());
D3D::context->OMSetRenderTargets(1, &FramebufferManager::GetEFBColorTexture()->GetRTV(), FramebufferManager::GetEFBDepthTexture()->GetDSV());
g_renderer->RestoreAPIState();
}
void MeshRenderer::Init(int numVertices, const MeshVertex* vertices, const MeshColor* color, const MeshSkin* skin, int numIndices, const AFIndex* indices)
{
Destroy();
static InputElement layout[] = {
CInputElement("COLOR", SF_R8G8B8A8_UNORM, 0),
CInputElement("TEXCOORD", SF_R32G32_FLOAT, 4),
};
shaderId = shaderMan.Create("skin", layout, dimof(layout), BM_NONE, DSM_DEPTH_ENABLE, CM_CW);
deviceMan11.GetDevice()->CreateBuffer(&CD3D11_BUFFER_DESC(numVertices * sizeof(MeshVertex), D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_UNORDERED_ACCESS, D3D11_USAGE_DEFAULT, 0, D3D11_RESOURCE_MISC_BUFFER_STRUCTURED, sizeof(MeshVertex)), nullptr, &skinnedPosBuffer);
D3D11_SUBRESOURCE_DATA subresData = { vertices, 0, 0 };
deviceMan11.GetDevice()->CreateBuffer(&CD3D11_BUFFER_DESC(numVertices * sizeof(MeshVertex), D3D11_BIND_SHADER_RESOURCE, D3D11_USAGE_DEFAULT, 0, D3D11_RESOURCE_MISC_BUFFER_STRUCTURED, sizeof(MeshVertex)), &subresData, &posBuffer);
subresData.pSysMem = skin;
HRESULT hr = deviceMan11.GetDevice()->CreateBuffer(&CD3D11_BUFFER_DESC(numVertices * sizeof(MeshSkin), D3D11_BIND_SHADER_RESOURCE, D3D11_USAGE_DEFAULT, 0, D3D11_RESOURCE_MISC_BUFFER_STRUCTURED, sizeof(MeshSkin)), &subresData, &skinBuffer);
colorBuffer = afCreateVertexBuffer(numVertices * sizeof(MeshColor), color);
iboId = afCreateIndexBuffer(indices, numIndices);
uboId = afCreateUBO(sizeof(MeshConstantBuffer));
int strides[] = {sizeof(MeshColor)};
VBOID vbos[] = {colorBuffer};
vao = afCreateVAO(layout, dimof(layout), 1, vbos, strides, iboId);
sampler = afCreateSampler(SF_MIPMAP, SW_REPEAT);
}
void LineGeometryShader::Init()
{
m_ready = false;
HRESULT hr;
// Create constant buffer for uploading data to geometry shader
D3D11_BUFFER_DESC bd = CD3D11_BUFFER_DESC(sizeof(LineGSParams_Padded),
D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
hr = D3D::device->CreateBuffer(&bd, nullptr, &m_paramsBuffer);
CHECK(SUCCEEDED(hr), "create line geometry shader params buffer");
D3D::SetDebugObjectName(m_paramsBuffer, "line geometry shader params buffer");
m_ready = true;
}
void PSTextureEncoder::Init()
{
// TODO: Move this to a constant somewhere in common.
TextureConfig encoding_texture_config(EFB_WIDTH * 4, 1024, 1, 1, AbstractTextureFormat::BGRA8,
true);
m_encoding_render_texture = g_renderer->CreateTexture(encoding_texture_config);
m_encoding_readback_texture =
g_renderer->CreateStagingTexture(StagingTextureType::Readback, encoding_texture_config);
_assert_(m_encoding_render_texture && m_encoding_readback_texture);
// Create constant buffer for uploading data to shaders
D3D11_BUFFER_DESC bd = CD3D11_BUFFER_DESC(sizeof(EFBEncodeParams), D3D11_BIND_CONSTANT_BUFFER);
HRESULT hr = D3D::device->CreateBuffer(&bd, nullptr, &m_encode_params);
CHECK(SUCCEEDED(hr), "create efb encode params buffer");
D3D::SetDebugObjectName(m_encode_params, "efb encoder params buffer");
}
void PixelShaderCache::Init()
{
unsigned int cbsize = ((sizeof(PixelShaderConstants))&(~0xf))+0x10; // must be a multiple of 16
D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(cbsize, D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
D3D::device->CreateBuffer(&cbdesc, NULL, &pscbuf);
CHECK(pscbuf!=NULL, "Create pixel shader constant buffer");
D3D::SetDebugObjectName((ID3D11DeviceChild*)pscbuf, "pixel shader constant buffer used to emulate the GX pipeline");
// used when drawing clear quads
s_ClearProgram = D3D::CompileAndCreatePixelShader(clear_program_code, sizeof(clear_program_code));
CHECK(s_ClearProgram!=NULL, "Create clear pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_ClearProgram, "clear pixel shader");
// used when copying/resolving the color buffer
s_ColorCopyProgram[0] = D3D::CompileAndCreatePixelShader(color_copy_program_code, sizeof(color_copy_program_code));
CHECK(s_ColorCopyProgram[0]!=NULL, "Create color copy pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_ColorCopyProgram[0], "color copy pixel shader");
// used for color conversion
s_ColorMatrixProgram[0] = D3D::CompileAndCreatePixelShader(color_matrix_program_code, sizeof(color_matrix_program_code));
CHECK(s_ColorMatrixProgram[0]!=NULL, "Create color matrix pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_ColorMatrixProgram[0], "color matrix pixel shader");
// used for depth copy
s_DepthMatrixProgram[0] = D3D::CompileAndCreatePixelShader(depth_matrix_program, sizeof(depth_matrix_program));
CHECK(s_DepthMatrixProgram[0]!=NULL, "Create depth matrix pixel shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)s_DepthMatrixProgram[0], "depth matrix pixel shader");
Clear();
if (!File::Exists(File::GetUserPath(D_SHADERCACHE_IDX)))
File::CreateDir(File::GetUserPath(D_SHADERCACHE_IDX).c_str());
SETSTAT(stats.numPixelShadersCreated, 0);
SETSTAT(stats.numPixelShadersAlive, 0);
char cache_filename[MAX_PATH];
sprintf(cache_filename, "%sdx11-%s-ps.cache", File::GetUserPath(D_SHADERCACHE_IDX).c_str(),
SConfig::GetInstance().m_LocalCoreStartupParameter.m_strUniqueID.c_str());
PixelShaderCacheInserter inserter;
g_ps_disk_cache.OpenAndRead(cache_filename, inserter);
if (g_Config.bEnableShaderDebugging)
Clear();
last_entry = NULL;
}
void VertexManager::CreateDeviceObjects()
{
D3D11_BUFFER_DESC bufdesc = CD3D11_BUFFER_DESC(MAX_BUFFER_SIZE,
D3D11_BIND_INDEX_BUFFER | D3D11_BIND_VERTEX_BUFFER, D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
m_vertexDrawOffset = 0;
m_indexDrawOffset = 0;
for (int i = 0; i < MAX_BUFFER_COUNT; i++)
{
m_buffers[i] = nullptr;
CHECK(SUCCEEDED(D3D::device->CreateBuffer(&bufdesc, nullptr, D3D::ToAddr(m_buffers[i]))), "Failed to create buffer.");
D3D::SetDebugObjectName((ID3D11DeviceChild*)m_buffers[i].get(), "Buffer of VertexManager");
}
m_currentBuffer = 0;
m_bufferCursor = MAX_BUFFER_SIZE;
}
void PSTextureEncoder::Init()
{
m_ready = false;
HRESULT hr;
// Create output texture RGBA format
D3D11_TEXTURE2D_DESC t2dd = CD3D11_TEXTURE2D_DESC(
DXGI_FORMAT_B8G8R8A8_UNORM,
EFB_WIDTH * 4, EFB_HEIGHT / 4, 1, 1, D3D11_BIND_RENDER_TARGET);
hr = D3D::device->CreateTexture2D(&t2dd, nullptr, &m_out);
CHECK(SUCCEEDED(hr), "create efb encode output texture");
D3D::SetDebugObjectName(m_out, "efb encoder output texture");
// Create output render target view
D3D11_RENDER_TARGET_VIEW_DESC rtvd = CD3D11_RENDER_TARGET_VIEW_DESC(m_out,
D3D11_RTV_DIMENSION_TEXTURE2D, DXGI_FORMAT_B8G8R8A8_UNORM);
hr = D3D::device->CreateRenderTargetView(m_out, &rtvd, &m_outRTV);
CHECK(SUCCEEDED(hr), "create efb encode output render target view");
D3D::SetDebugObjectName(m_outRTV, "efb encoder output rtv");
// Create output staging buffer
t2dd.Usage = D3D11_USAGE_STAGING;
t2dd.BindFlags = 0;
t2dd.CPUAccessFlags = D3D11_CPU_ACCESS_READ;
hr = D3D::device->CreateTexture2D(&t2dd, nullptr, &m_outStage);
CHECK(SUCCEEDED(hr), "create efb encode output staging buffer");
D3D::SetDebugObjectName(m_outStage, "efb encoder output staging buffer");
// Create constant buffer for uploading data to shaders
D3D11_BUFFER_DESC bd = CD3D11_BUFFER_DESC(sizeof(EFBEncodeParams),
D3D11_BIND_CONSTANT_BUFFER);
hr = D3D::device->CreateBuffer(&bd, nullptr, &m_encodeParams);
CHECK(SUCCEEDED(hr), "create efb encode params buffer");
D3D::SetDebugObjectName(m_encodeParams, "efb encoder params buffer");
m_ready = true;
}
void BBox::Init()
{
if (g_ActiveConfig.backend_info.bSupportsBBox)
{
// Create 2 buffers here.
// First on Default pool.
auto desc = CD3D11_BUFFER_DESC(4 * sizeof(s32), D3D11_BIND_UNORDERED_ACCESS, D3D11_USAGE_DEFAULT, 0, 0, 4);
s_values[0] = s_values[1] = s_values[2] = s_values[3] = 0;
D3D11_SUBRESOURCE_DATA data;
data.pSysMem = s_values;
data.SysMemPitch = 4 * sizeof(s32);
data.SysMemSlicePitch = 0;
HRESULT hr;
hr = D3D::device->CreateBuffer(&desc, &data, ToAddr(s_bbox_buffer));
CHECK(SUCCEEDED(hr), "Create BoundingBox buffer.");
D3D::SetDebugObjectName(s_bbox_buffer.get(), "BoundingBox buffer");
// Second to use as a staging buffer.
desc.Usage = D3D11_USAGE_STAGING;
desc.CPUAccessFlags = D3D11_CPU_ACCESS_READ;
desc.BindFlags = 0;
hr = D3D::device->CreateBuffer(&desc, nullptr, ToAddr(s_bbox_staging_buffer));
CHECK(SUCCEEDED(hr), "Create BoundingBox staging buffer.");
D3D::SetDebugObjectName(s_bbox_staging_buffer.get(), "BoundingBox staging buffer");
// UAV is required to allow concurrent access.
D3D11_UNORDERED_ACCESS_VIEW_DESC UAVdesc = {};
UAVdesc.Format = DXGI_FORMAT_R32_SINT;
UAVdesc.ViewDimension = D3D11_UAV_DIMENSION_BUFFER;
UAVdesc.Buffer.FirstElement = 0;
UAVdesc.Buffer.Flags = 0;
UAVdesc.Buffer.NumElements = 4;
hr = D3D::device->CreateUnorderedAccessView(s_bbox_buffer.get(), &UAVdesc, ToAddr(s_bbox_uav));
CHECK(SUCCEEDED(hr), "Create BoundingBox UAV.");
s_cpu_dirty = true;
s_gpu_dirty = true;
}
}
//////////////////////////////////////////////////////////////////////////
// d3d11 constant buffer
cgl::CD3D11ConstantBuffer::CD3D11ConstantBuffer( UINT elementSize, D3D11_USAGE usage /*= D3D11_USAGE_DEFAULT*/, UINT cpuAccessFlags /*= 0*/, UINT miscFlags /*= 0*/ )
: CD3D11Buffer(elementSize, CD3D11_BUFFER_DESC(0, D3D11_BIND_CONSTANT_BUFFER, usage, cpuAccessFlags, miscFlags, elementSize ))
{
}
//////////////////////////////////////////////////////////////////////////
// d3d11 index buffer
cgl::CD3D11IndexBuffer::CD3D11IndexBuffer( UINT elementSize, D3D11_USAGE usage /*= D3D11_USAGE_DEFAULT*/, UINT cpuAccessFlags /*= 0*/, UINT miscFlags /*= 0*/ )
: CD3D11Buffer(elementSize, CD3D11_BUFFER_DESC(0, D3D11_BIND_INDEX_BUFFER, usage, cpuAccessFlags, miscFlags, elementSize ))
{
}
//////////////////////////////////////////////////////////////////////////
// d3d11 vertex buffer
cgl::CD3D11VertexBuffer::CD3D11VertexBuffer(UINT stride, D3D11_USAGE usage, UINT cpuAccessFlags, UINT miscFlags )
: CD3D11Buffer(stride, CD3D11_BUFFER_DESC(0, D3D11_BIND_VERTEX_BUFFER, usage, cpuAccessFlags, miscFlags ))
{
}
void TextureCache::TCacheEntry::FromRenderTarget(u32 dstAddr, unsigned int dstFormat,
PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool isIntensity, bool scaleByHalf, unsigned int cbufid,
const float *colmat)
{
if (type != TCET_EC_DYNAMIC || g_ActiveConfig.bCopyEFBToTexture)
{
g_renderer->ResetAPIState();
// stretch picture with increased internal resolution
const D3D11_VIEWPORT vp = CD3D11_VIEWPORT(0.f, 0.f, (float)virtual_width, (float)virtual_height);
D3D::context->RSSetViewports(1, &vp);
// set transformation
if (nullptr == efbcopycbuf[cbufid])
{
const D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(28 * sizeof(float), D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DEFAULT);
D3D11_SUBRESOURCE_DATA data;
data.pSysMem = colmat;
HRESULT hr = D3D::device->CreateBuffer(&cbdesc, &data, &efbcopycbuf[cbufid]);
CHECK(SUCCEEDED(hr), "Create efb copy constant buffer %d", cbufid);
D3D::SetDebugObjectName((ID3D11DeviceChild*)efbcopycbuf[cbufid], "a constant buffer used in TextureCache::CopyRenderTargetToTexture");
}
D3D::stateman->SetPixelConstants(efbcopycbuf[cbufid]);
const TargetRectangle targetSource = g_renderer->ConvertEFBRectangle(srcRect);
// TODO: try targetSource.asRECT();
const D3D11_RECT sourcerect = CD3D11_RECT(targetSource.left, targetSource.top, targetSource.right, targetSource.bottom);
// Use linear filtering if (bScaleByHalf), use point filtering otherwise
if (scaleByHalf)
D3D::SetLinearCopySampler();
else
D3D::SetPointCopySampler();
// if texture is currently in use, it needs to be temporarily unset
u32 textureSlotMask = D3D::stateman->UnsetTexture(texture->GetSRV());
D3D::stateman->Apply();
D3D::context->OMSetRenderTargets(1, &texture->GetRTV(), nullptr);
// Create texture copy
D3D::drawShadedTexQuad(
(srcFormat == PEControl::Z24) ? FramebufferManager::GetEFBDepthTexture()->GetSRV() : FramebufferManager::GetEFBColorTexture()->GetSRV(),
&sourcerect, Renderer::GetTargetWidth(), Renderer::GetTargetHeight(),
(srcFormat == PEControl::Z24) ? PixelShaderCache::GetDepthMatrixProgram(true) : PixelShaderCache::GetColorMatrixProgram(true),
VertexShaderCache::GetSimpleVertexShader(), VertexShaderCache::GetSimpleInputLayout(), GeometryShaderCache::GetCopyGeometryShader());
D3D::context->OMSetRenderTargets(1, &FramebufferManager::GetEFBColorTexture()->GetRTV(), FramebufferManager::GetEFBDepthTexture()->GetDSV());
g_renderer->RestoreAPIState();
// Restore old texture in all previously used slots, if any
D3D::stateman->SetTextureByMask(textureSlotMask, texture->GetSRV());
}
if (!g_ActiveConfig.bCopyEFBToTexture)
{
u8* dst = Memory::GetPointer(dstAddr);
size_t encoded_size = g_encoder->Encode(dst, dstFormat, srcFormat, srcRect, isIntensity, scaleByHalf);
u64 hash = GetHash64(dst, (int)encoded_size, g_ActiveConfig.iSafeTextureCache_ColorSamples);
// Mark texture entries in destination address range dynamic unless caching is enabled and the texture entry is up to date
if (!g_ActiveConfig.bEFBCopyCacheEnable)
TextureCache::MakeRangeDynamic(addr, (u32)encoded_size);
else if (!TextureCache::Find(addr, hash))
TextureCache::MakeRangeDynamic(addr, (u32)encoded_size);
this->hash = hash;
}
}
//////////////////////////////////////////////////////////////////////////
// d3d11 input buffer
cgl::util::CGLInputBuffer::CGLInputBuffer(cgl::core::PD3D11InputLayout pLayout, ICGLInputBufferDataProvider* pDataProvider, D3D11_USAGE usage /*= D3D11_USAGE_DEFAULT*/, DWORD cpuAccessFlags /*= 0*/)
: CGLBase("CD3D11InputBuffer"), CGLTypedBindable(this), m_pLayout(pLayout), m_pDataProvider(pDataProvider), m_desc(CD3D11_BUFFER_DESC(0, D3D11_BIND_VERTEX_BUFFER, usage, cpuAccessFlags))
{
SetParam<UINT>(BIND_PARAM_STRIDE_UINT, 1);
SetParam<UINT>(BIND_PARAM_OFFSET_UINT, 0);
SetParam<UINT>(BIND_PARAM_SLOT_UINT, 0);
}
void VertexShaderCache::Init()
{
const D3D11_INPUT_ELEMENT_DESC simpleelems[2] =
{
{ "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "TEXCOORD", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0 },
};
const D3D11_INPUT_ELEMENT_DESC clearelems[2] =
{
{ "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0 },
};
unsigned int cbsize = ((sizeof(vsconstants))&(~0xf))+0x10; // must be a multiple of 16
D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(cbsize, D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
HRESULT hr = D3D::device->CreateBuffer(&cbdesc, NULL, &vscbuf);
CHECK(hr==S_OK, "Create vertex shader constant buffer (size=%u)", cbsize);
D3D::SetDebugObjectName((ID3D11DeviceChild*)vscbuf, "vertex shader constant buffer used to emulate the GX pipeline");
D3DBlob* blob;
D3D::CompileVertexShader(simple_shader_code, sizeof(simple_shader_code), &blob);
D3D::device->CreateInputLayout(simpleelems, 2, blob->Data(), blob->Size(), &SimpleLayout);
SimpleVertexShader = D3D::CreateVertexShaderFromByteCode(blob);
if (SimpleLayout == NULL || SimpleVertexShader == NULL) PanicAlert("Failed to create simple vertex shader or input layout at %s %d\n", __FILE__, __LINE__);
blob->Release();
D3D::SetDebugObjectName((ID3D11DeviceChild*)SimpleVertexShader, "simple vertex shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)SimpleLayout, "simple input layout");
D3D::CompileVertexShader(clear_shader_code, sizeof(clear_shader_code), &blob);
D3D::device->CreateInputLayout(clearelems, 2, blob->Data(), blob->Size(), &ClearLayout);
ClearVertexShader = D3D::CreateVertexShaderFromByteCode(blob);
if (ClearLayout == NULL || ClearVertexShader == NULL) PanicAlert("Failed to create clear vertex shader or input layout at %s %d\n", __FILE__, __LINE__);
blob->Release();
D3D::SetDebugObjectName((ID3D11DeviceChild*)ClearVertexShader, "clear vertex shader");
D3D::SetDebugObjectName((ID3D11DeviceChild*)ClearLayout, "clear input layout");
Clear();
// these values are hardcoded, they depend on internal D3DCompile behavior
// TODO: Do this with D3DReflect or something instead
unsigned int k;
for (k = 0;k < 6;k++) vs_constant_offset_table[C_POSNORMALMATRIX+k] = 0+4*k;
for (k = 0;k < 4;k++) vs_constant_offset_table[C_PROJECTION+k] = 24+4*k;
for (k = 0;k < 4;k++) vs_constant_offset_table[C_MATERIALS+k] = 40+4*k;
for (k = 0;k < 40;k++) vs_constant_offset_table[C_LIGHTS+k] = 56+4*k;
for (k = 0;k < 24;k++) vs_constant_offset_table[C_TEXMATRICES+k] = 216+4*k;
for (k = 0;k < 64;k++) vs_constant_offset_table[C_TRANSFORMMATRICES+k] = 312+4*k;
for (k = 0;k < 32;k++) vs_constant_offset_table[C_NORMALMATRICES+k] = 568+4*k;
for (k = 0;k < 64;k++) vs_constant_offset_table[C_POSTTRANSFORMMATRICES+k] = 696+4*k;
for (k = 0;k < 4;k++) vs_constant_offset_table[C_DEPTHPARAMS+k] = 952+4*k;
if (!File::Exists(File::GetUserPath(D_SHADERCACHE_IDX)))
File::CreateDir(File::GetUserPath(D_SHADERCACHE_IDX).c_str());
SETSTAT(stats.numVertexShadersCreated, 0);
SETSTAT(stats.numVertexShadersAlive, 0);
char cache_filename[MAX_PATH];
sprintf(cache_filename, "%sdx11-%s-vs.cache", File::GetUserPath(D_SHADERCACHE_IDX).c_str(),
SConfig::GetInstance().m_LocalCoreStartupParameter.m_strUniqueID.c_str());
VertexShaderCacheInserter inserter;
g_vs_disk_cache.OpenAndRead(cache_filename, inserter);
if (g_Config.bEnableShaderDebugging)
Clear();
last_entry = NULL;
}
void init( HWND hWnd )
{
HRESULT hr;
DXGI_SWAP_CHAIN_DESC scDesc;
memset( &scDesc, 0, sizeof scDesc );
scDesc.BufferDesc.Width = g_windowWidth;
scDesc.BufferDesc.Height = g_windowHeight;
scDesc.BufferDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM;
scDesc.BufferDesc.RefreshRate.Denominator = 1;
scDesc.BufferDesc.RefreshRate.Numerator = 60;
scDesc.BufferUsage = DXGI_USAGE_RENDER_TARGET_OUTPUT;
scDesc.BufferCount = 1;
scDesc.SwapEffect = DXGI_SWAP_EFFECT_DISCARD;
scDesc.SampleDesc.Count = 1;
scDesc.Windowed = TRUE;
scDesc.OutputWindow = hWnd;
D3D_FEATURE_LEVEL features[] = { D3D_FEATURE_LEVEL_11_0 };
UINT flags = D3D11_CREATE_DEVICE_BGRA_SUPPORT;
#ifndef NDEBUG
flags |= D3D11_CREATE_DEVICE_DEBUG;
#endif
IDXGISwapChain* swapChain;
ID3D11Device* device;
ID3D11DeviceContext* context;
hr = D3D11CreateDeviceAndSwapChain(
nullptr, D3D_DRIVER_TYPE_HARDWARE, NULL, flags, features, ARRAYSIZE( features ),
D3D11_SDK_VERSION, &scDesc, &swapChain, &device, &featureLevel_, &context );
Assert( hr );
swapChain_.reset( swapChain );
device_.reset( device );
context_.reset( context );
ID3D11Texture2D* backBuffer;
ID3D11RenderTargetView* backBufferRTV;
hr = swapChain->GetBuffer( 0, __uuidof( ID3D11Texture2D ), reinterpret_cast< void** >( &backBuffer ) );
Assert( hr );
hr = device_->CreateRenderTargetView( backBuffer, nullptr, &backBufferRTV );
Assert( hr );
backBufferRTV_.reset( backBufferRTV );
backBuffer->Release();
// Common state
ID3D11RasterizerState* rs;
D3D11_RASTERIZER_DESC rsDesc;
memset( &rsDesc, 0, sizeof rsDesc );
rsDesc.CullMode = D3D11_CULL_BACK;
rsDesc.FillMode = D3D11_FILL_SOLID;
rsDesc.DepthClipEnable = TRUE;
hr = device_->CreateRasterizerState( &rsDesc, &rs );
Assert( hr );
rasterState_.reset( rs );
// Body
std::string binData;
std::string vsBinData;
ID3D11VertexShader* vs;
binData = fileGetContents( "def.vs.cso" );
hr = device_->CreateVertexShader(
reinterpret_cast< const void* >( binData.c_str() ), binData.size(), nullptr, &vs );
Assert( hr );
modelVS_.reset( vs );
vsBinData = binData;
ID3D11PixelShader* ps;
binData = fileGetContents( "def.ps.cso" );
hr = device_->CreatePixelShader(
reinterpret_cast< const void* >( binData.c_str() ), binData.size(), nullptr, &ps );
Assert( hr );
modelPS_.reset( ps );
D3D11_BUFFER_DESC bufDesc = CD3D11_BUFFER_DESC( 64, D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DEFAULT, 0 );
ID3D11Buffer* buf;
hr = device_->CreateBuffer( &bufDesc, nullptr, &buf );
Assert( hr );
modelCB_.reset( buf );
// 四角錐ポリゴン
static const MeshFormat pyramidVertex[] = {
{ -1.0f, 0.0f, -1.0f, 0xFF33AAAA },
{ 1.0f, 0.0f, -1.0f, 0xFF33AAAA },
{ -1.0f, 0.0f, 1.0f, 0xFF33AAAA },
{ -1.0f, 0.0f, 1.0f, 0xFF33AAAA },
{ 1.0f, 0.0f, -1.0f, 0xFF33AAAA },
{ 1.0f, 0.0f, 1.0f, 0xFF33AAAA },
{ 1.0f, 0.0f, -1.0f, 0xFFEE33BB },
{ -1.0f, 0.0f, -1.0f, 0xFFEE33BB },
{ 0.0f, 1.0f, 0.0f, 0xFFEE33BB },
{ -1.0f, 0.0f, -1.0f, 0xFFCC33BB },
{ -1.0f, 0.0f, 1.0f, 0xFFCC33BB },
{ 0.0f, 1.0f, 0.0f, 0xFFCC33BB },
{ -1.0f, 0.0f, 1.0f, 0xFFAA33BB },
{ 1.0f, 0.0f, 1.0f, 0xFFAA33BB },
{ 0.0f, 1.0f, 0.0f, 0xFFAA33BB },
{ 1.0f, 0.0f, 1.0f, 0xFF8833BB },
{ 1.0f, 0.0f, -1.0f, 0xFF8833BB },
{ 0.0f, 1.0f, 0.0f, 0xFF8833BB },
};
bufDesc = CD3D11_BUFFER_DESC( sizeof pyramidVertex, D3D11_BIND_VERTEX_BUFFER, D3D11_USAGE_IMMUTABLE, 0 );
D3D11_SUBRESOURCE_DATA subresData;
subresData.pSysMem = pyramidVertex;
subresData.SysMemPitch = subresData.SysMemSlicePitch = 0;
hr = device_->CreateBuffer( &bufDesc, &subresData, &buf );
Assert( hr );
modelVB_.reset( buf );
D3D11_INPUT_ELEMENT_DESC ieDesc[] = {
{ "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "COLOR", 0, DXGI_FORMAT_B8G8R8A8_UNORM, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0 }
};
ID3D11InputLayout* il;
hr = device_->CreateInputLayout( ieDesc, ARRAYSIZE( ieDesc ), vsBinData.data(), vsBinData.size(), &il );
Assert( hr );
modelIL_.reset( il );
}
void XFBEncoder::Init()
{
HRESULT hr;
// Create output texture
// The pixel shader can generate one YUYV entry per pixel. One YUYV entry
// is created for every two EFB pixels.
D3D11_TEXTURE2D_DESC t2dd = CD3D11_TEXTURE2D_DESC(
DXGI_FORMAT_R8G8B8A8_UNORM, MAX_XFB_WIDTH/2, MAX_XFB_HEIGHT, 1, 1,
D3D11_BIND_RENDER_TARGET);
hr = D3D::device->CreateTexture2D(&t2dd, nullptr, &m_out);
CHECK(SUCCEEDED(hr), "create xfb encoder output texture");
D3D::SetDebugObjectName(m_out, "xfb encoder output texture");
// Create output render target view
D3D11_RENDER_TARGET_VIEW_DESC rtvd = CD3D11_RENDER_TARGET_VIEW_DESC(m_out,
D3D11_RTV_DIMENSION_TEXTURE2D, DXGI_FORMAT_R8G8B8A8_UNORM);
hr = D3D::device->CreateRenderTargetView(m_out, &rtvd, &m_outRTV);
CHECK(SUCCEEDED(hr), "create xfb encoder output texture rtv");
D3D::SetDebugObjectName(m_outRTV, "xfb encoder output rtv");
// Create output staging buffer
t2dd.Usage = D3D11_USAGE_STAGING;
t2dd.BindFlags = 0;
t2dd.CPUAccessFlags = D3D11_CPU_ACCESS_READ;
hr = D3D::device->CreateTexture2D(&t2dd, nullptr, &m_outStage);
CHECK(SUCCEEDED(hr), "create xfb encoder output staging buffer");
D3D::SetDebugObjectName(m_outStage, "xfb encoder output staging buffer");
// Create constant buffer for uploading params to shaders
D3D11_BUFFER_DESC bd = CD3D11_BUFFER_DESC(sizeof(XFBEncodeParams),
D3D11_BIND_CONSTANT_BUFFER);
hr = D3D::device->CreateBuffer(&bd, nullptr, &m_encodeParams);
CHECK(SUCCEEDED(hr), "create xfb encode params buffer");
D3D::SetDebugObjectName(m_encodeParams, "xfb encoder params buffer");
// Create vertex quad
bd = CD3D11_BUFFER_DESC(sizeof(QUAD_VERTS), D3D11_BIND_VERTEX_BUFFER,
D3D11_USAGE_IMMUTABLE);
D3D11_SUBRESOURCE_DATA srd = { QUAD_VERTS, 0, 0 };
hr = D3D::device->CreateBuffer(&bd, &srd, &m_quad);
CHECK(SUCCEEDED(hr), "create xfb encode quad vertex buffer");
D3D::SetDebugObjectName(m_quad, "xfb encoder quad vertex buffer");
// Create vertex shader
D3DBlob* bytecode = nullptr;
if (!D3D::CompileVertexShader(XFB_ENCODE_VS, &bytecode))
{
ERROR_LOG(VIDEO, "XFB encode vertex shader failed to compile");
return;
}
hr = D3D::device->CreateVertexShader(bytecode->Data(), bytecode->Size(), nullptr, &m_vShader);
CHECK(SUCCEEDED(hr), "create xfb encode vertex shader");
D3D::SetDebugObjectName(m_vShader, "xfb encoder vertex shader");
// Create input layout for vertex quad using bytecode from vertex shader
hr = D3D::device->CreateInputLayout(QUAD_LAYOUT_DESC,
sizeof(QUAD_LAYOUT_DESC)/sizeof(D3D11_INPUT_ELEMENT_DESC),
bytecode->Data(), bytecode->Size(), &m_quadLayout);
CHECK(SUCCEEDED(hr), "create xfb encode quad vertex layout");
D3D::SetDebugObjectName(m_quadLayout, "xfb encoder quad layout");
bytecode->Release();
// Create pixel shader
m_pShader = D3D::CompileAndCreatePixelShader(XFB_ENCODE_PS);
if (!m_pShader)
{
ERROR_LOG(VIDEO, "XFB encode pixel shader failed to compile");
return;
}
D3D::SetDebugObjectName(m_pShader, "xfb encoder pixel shader");
// Create blend state
D3D11_BLEND_DESC bld = CD3D11_BLEND_DESC(CD3D11_DEFAULT());
hr = D3D::device->CreateBlendState(&bld, &m_xfbEncodeBlendState);
CHECK(SUCCEEDED(hr), "create xfb encode blend state");
D3D::SetDebugObjectName(m_xfbEncodeBlendState, "xfb encoder blend state");
// Create depth state
D3D11_DEPTH_STENCIL_DESC dsd = CD3D11_DEPTH_STENCIL_DESC(CD3D11_DEFAULT());
dsd.DepthEnable = FALSE;
hr = D3D::device->CreateDepthStencilState(&dsd, &m_xfbEncodeDepthState);
CHECK(SUCCEEDED(hr), "create xfb encode depth state");
D3D::SetDebugObjectName(m_xfbEncodeDepthState, "xfb encoder depth state");
// Create rasterizer state
D3D11_RASTERIZER_DESC rd = CD3D11_RASTERIZER_DESC(CD3D11_DEFAULT());
rd.CullMode = D3D11_CULL_NONE;
rd.DepthClipEnable = FALSE;
hr = D3D::device->CreateRasterizerState(&rd, &m_xfbEncodeRastState);
CHECK(SUCCEEDED(hr), "create xfb encode rasterizer state");
D3D::SetDebugObjectName(m_xfbEncodeRastState, "xfb encoder rast state");
// Create EFB texture sampler
D3D11_SAMPLER_DESC sd = CD3D11_SAMPLER_DESC(CD3D11_DEFAULT());
// FIXME: Should we really use point sampling here?
sd.Filter = D3D11_FILTER_MIN_MAG_MIP_POINT;
hr = D3D::device->CreateSamplerState(&sd, &m_efbSampler);
CHECK(SUCCEEDED(hr), "create xfb encode texture sampler");
D3D::SetDebugObjectName(m_efbSampler, "xfb encoder texture sampler");
}
OpenGL::OpenGL(ID3D11Device* device, ID3D11DeviceContext* context)
{
s_instance = this;
_orthoLeft = 0;
_orthoRight = 0;
_orthoBottom = 0;
_orthoTop = 0;
_device = device;
_context = context;
_vertexBuffer = NULL;
_currentTextureId = 0;
_enableTexture2D = false;
_batchCurrentTextureId = 0;
_batchEnableTexture2D = false;
_loadingComplete = false;
_vertexBuilder = new VertexBuilder();
_maxVertices = MaxBatchVertices;
_firstDrawInFrame = false;
_currentBatchVertex = 0;
_enableScissoring = false;
_scissorRect.left = 0;
_scissorRect.right = 0;
_scissorRect.top = 0;
_scissorRect.bottom = 0;
_vertexShader = NULL;
_pixelShaderTexture = NULL;
_pixelShaderColor = NULL;
auto vsize = sizeof (this->_batchVertices);
// Asynchronously load the vertex and pixel shaders
auto loadVSTask = ReadDataAsync("OpenGLVS.cso");
auto loadPSColorTask = ReadDataAsync("OpenGLColorPS.cso");
auto loadPSTextureTask = ReadDataAsync("OpenGLTexturePS.cso");
auto createVSTask = loadVSTask.then([this](ByteArray ba) {
auto bytecodeVS = ba.data;
ThrowIfFailed(
_device->CreateVertexShader(
bytecodeVS->Data,
bytecodeVS->Length,
nullptr,
&_vertexShader
)
);
const D3D11_INPUT_ELEMENT_DESC vertexDesc[] =
{
{ "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "COLOR", 0, DXGI_FORMAT_R32G32B32A32_FLOAT, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0 },
};
ThrowIfFailed(
_device->CreateInputLayout(
vertexDesc,
ARRAYSIZE(vertexDesc),
bytecodeVS->Data,
bytecodeVS->Length,
&_inputLayout
)
);
});
auto createPSColorTask = loadPSColorTask.then([this](ByteArray ba) {
auto bytecodePS = ba.data;
ThrowIfFailed(
_device->CreatePixelShader(
bytecodePS->Data,
bytecodePS->Length,
nullptr,
&_pixelShaderColor
)
);
});
auto createPSTextureTask = loadPSTextureTask.then([this](ByteArray ba) {
auto bytecodePS = ba.data;
ThrowIfFailed(
_device->CreatePixelShader(
bytecodePS->Data,
bytecodePS->Length,
nullptr,
&_pixelShaderTexture
)
);
ThrowIfFailed(
_device->CreateBuffer(
&CD3D11_BUFFER_DESC(sizeof(ModelViewProjectionConstantBuffer), D3D11_BIND_CONSTANT_BUFFER),
nullptr,
&_constantBuffer
));
});
createPSTextureTask.then([this] () {
_loadingComplete = true;
});
ZeroMemory(&_rasterizerDesc, sizeof(_rasterizerDesc));
_rasterizerDesc.FillMode = D3D11_FILL_MODE::D3D11_FILL_SOLID;
_rasterizerDesc.CullMode = D3D11_CULL_MODE::D3D11_CULL_NONE;
_rasterizerDesc.FrontCounterClockwise = true;
_rasterizerDesc.DepthBias = 0;
_rasterizerDesc.SlopeScaledDepthBias = 0.f;
_rasterizerDesc.DepthBiasClamp = 0.f;
_rasterizerDesc.DepthClipEnable = true;
_rasterizerDesc.ScissorEnable = false;
_rasterizerDesc.MultisampleEnable = false;
_rasterizerDesc.AntialiasedLineEnable = false;
ID3D11RasterizerState* rasterizerState;
_device->CreateRasterizerState(&_rasterizerDesc, &rasterizerState);
_context->RSSetState(rasterizerState);
// Push an identity matrix on the stack to start with
DirectX::XMFLOAT4X4* matrix = new DirectX::XMFLOAT4X4();
auto identity = DirectX::XMMatrixIdentity();
DirectX::XMStoreFloat4x4(matrix, identity);
_matrices.push(matrix);
}
namespace DX11
{
static std::unique_ptr<TextureEncoder> g_encoder;
const size_t MAX_COPY_BUFFERS = 32;
ID3D11Buffer* efbcopycbuf[MAX_COPY_BUFFERS] = { 0 };
TextureCache::TCacheEntry::~TCacheEntry()
{
texture->Release();
}
void TextureCache::TCacheEntry::Bind(unsigned int stage)
{
D3D::stateman->SetTexture(stage, texture->GetSRV());
}
bool TextureCache::TCacheEntry::Save(const std::string& filename, unsigned int level)
{
// TODO: Somehow implement this (D3DX11 doesn't support dumping individual LODs)
static bool warn_once = true;
if (level && warn_once)
{
WARN_LOG(VIDEO, "Dumping individual LOD not supported by D3D11 backend!");
warn_once = false;
return false;
}
ID3D11Texture2D* pNewTexture = nullptr;
ID3D11Texture2D* pSurface = texture->GetTex();
D3D11_TEXTURE2D_DESC desc;
pSurface->GetDesc(&desc);
desc.BindFlags = 0;
desc.CPUAccessFlags = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE;
desc.Usage = D3D11_USAGE_STAGING;
HRESULT hr = D3D::device->CreateTexture2D(&desc, nullptr, &pNewTexture);
bool saved_png = false;
if (SUCCEEDED(hr) && pNewTexture)
{
D3D::context->CopyResource(pNewTexture, pSurface);
D3D11_MAPPED_SUBRESOURCE map;
hr = D3D::context->Map(pNewTexture, 0, D3D11_MAP_READ_WRITE, 0, &map);
if (SUCCEEDED(hr))
{
saved_png = TextureToPng((u8*)map.pData, map.RowPitch, filename, desc.Width, desc.Height);
D3D::context->Unmap(pNewTexture, 0);
}
SAFE_RELEASE(pNewTexture);
}
return saved_png;
}
void TextureCache::TCacheEntry::CopyRectangleFromTexture(
const TCacheEntryBase* source,
const MathUtil::Rectangle<int> &srcrect,
const MathUtil::Rectangle<int> &dstrect)
{
TCacheEntry* srcentry = (TCacheEntry*)source;
if (srcrect.GetWidth() == dstrect.GetWidth()
&& srcrect.GetHeight() == dstrect.GetHeight())
{
const D3D11_BOX *psrcbox = nullptr;
D3D11_BOX srcbox;
if (srcrect.left != 0 || srcrect.top != 0)
{
srcbox.left = srcrect.left;
srcbox.top = srcrect.top;
srcbox.right = srcrect.right;
srcbox.bottom = srcrect.bottom;
psrcbox = &srcbox;
}
D3D::context->CopySubresourceRegion(
texture->GetTex(),
0,
dstrect.left,
dstrect.top,
0,
srcentry->texture->GetTex(),
0,
psrcbox);
return;
}
else if (!config.rendertarget)
{
return;
}
g_renderer->ResetAPIState(); // reset any game specific settings
const D3D11_VIEWPORT vp = CD3D11_VIEWPORT(
float(dstrect.left),
float(dstrect.top),
float(dstrect.GetWidth()),
float(dstrect.GetHeight()));
D3D::context->OMSetRenderTargets(1, &texture->GetRTV(), nullptr);
D3D::context->RSSetViewports(1, &vp);
D3D::SetLinearCopySampler();
D3D11_RECT srcRC;
srcRC.left = srcrect.left;
srcRC.right = srcrect.right;
srcRC.top = srcrect.top;
srcRC.bottom = srcrect.bottom;
D3D::drawShadedTexQuad(srcentry->texture->GetSRV(), &srcRC,
srcentry->config.width, srcentry->config.height,
PixelShaderCache::GetColorCopyProgram(false),
VertexShaderCache::GetSimpleVertexShader(),
VertexShaderCache::GetSimpleInputLayout(), nullptr, 1.0, 0);
D3D::context->OMSetRenderTargets(1,
&FramebufferManager::GetEFBColorTexture()->GetRTV(),
FramebufferManager::GetEFBDepthTexture()->GetDSV());
g_renderer->RestoreAPIState();
}
void TextureCache::TCacheEntry::Load(unsigned int width, unsigned int height,
unsigned int expanded_width, unsigned int level)
{
unsigned int src_pitch = 4 * expanded_width;
D3D::ReplaceRGBATexture2D(texture->GetTex(), TextureCache::temp, width, height, src_pitch, level, usage);
}
TextureCacheBase::TCacheEntryBase* TextureCache::CreateTexture(const TCacheEntryConfig& config)
{
if (config.rendertarget)
{
return new TCacheEntry(config, D3DTexture2D::Create(config.width, config.height,
(D3D11_BIND_FLAG)((int)D3D11_BIND_RENDER_TARGET | (int)D3D11_BIND_SHADER_RESOURCE),
D3D11_USAGE_DEFAULT, DXGI_FORMAT_R8G8B8A8_UNORM, 1, config.layers));
}
else
{
D3D11_USAGE usage = D3D11_USAGE_DEFAULT;
D3D11_CPU_ACCESS_FLAG cpu_access = (D3D11_CPU_ACCESS_FLAG)0;
if (config.levels == 1)
{
usage = D3D11_USAGE_DYNAMIC;
cpu_access = D3D11_CPU_ACCESS_WRITE;
}
const D3D11_TEXTURE2D_DESC texdesc = CD3D11_TEXTURE2D_DESC(DXGI_FORMAT_R8G8B8A8_UNORM,
config.width, config.height, 1, config.levels, D3D11_BIND_SHADER_RESOURCE, usage, cpu_access);
ID3D11Texture2D *pTexture;
const HRESULT hr = D3D::device->CreateTexture2D(&texdesc, nullptr, &pTexture);
CHECK(SUCCEEDED(hr), "Create texture of the TextureCache");
TCacheEntry* const entry = new TCacheEntry(config, new D3DTexture2D(pTexture, D3D11_BIND_SHADER_RESOURCE));
entry->usage = usage;
// TODO: better debug names
D3D::SetDebugObjectName((ID3D11DeviceChild*)entry->texture->GetTex(), "a texture of the TextureCache");
D3D::SetDebugObjectName((ID3D11DeviceChild*)entry->texture->GetSRV(), "shader resource view of a texture of the TextureCache");
SAFE_RELEASE(pTexture);
return entry;
}
}
void TextureCache::TCacheEntry::FromRenderTarget(u8* dst, PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool scaleByHalf, unsigned int cbufid, const float *colmat)
{
g_renderer->ResetAPIState();
// stretch picture with increased internal resolution
const D3D11_VIEWPORT vp = CD3D11_VIEWPORT(0.f, 0.f, (float)config.width, (float)config.height);
D3D::context->RSSetViewports(1, &vp);
// set transformation
if (nullptr == efbcopycbuf[cbufid])
{
const D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(28 * sizeof(float), D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DEFAULT);
D3D11_SUBRESOURCE_DATA data;
data.pSysMem = colmat;
HRESULT hr = D3D::device->CreateBuffer(&cbdesc, &data, &efbcopycbuf[cbufid]);
CHECK(SUCCEEDED(hr), "Create efb copy constant buffer %d", cbufid);
D3D::SetDebugObjectName((ID3D11DeviceChild*)efbcopycbuf[cbufid], "a constant buffer used in TextureCache::CopyRenderTargetToTexture");
}
D3D::stateman->SetPixelConstants(efbcopycbuf[cbufid]);
const TargetRectangle targetSource = g_renderer->ConvertEFBRectangle(srcRect);
// TODO: try targetSource.asRECT();
const D3D11_RECT sourcerect = CD3D11_RECT(targetSource.left, targetSource.top, targetSource.right, targetSource.bottom);
// Use linear filtering if (bScaleByHalf), use point filtering otherwise
if (scaleByHalf)
D3D::SetLinearCopySampler();
else
D3D::SetPointCopySampler();
// Make sure we don't draw with the texture set as both a source and target.
// (This can happen because we don't unbind textures when we free them.)
D3D::stateman->UnsetTexture(texture->GetSRV());
D3D::context->OMSetRenderTargets(1, &texture->GetRTV(), nullptr);
// Create texture copy
D3D::drawShadedTexQuad(
(srcFormat == PEControl::Z24 ? FramebufferManager::GetEFBDepthTexture() : FramebufferManager::GetEFBColorTexture())->GetSRV(),
&sourcerect, Renderer::GetTargetWidth(),
Renderer::GetTargetHeight(),
srcFormat == PEControl::Z24 ? PixelShaderCache::GetDepthMatrixProgram(true) : PixelShaderCache::GetColorMatrixProgram(true),
VertexShaderCache::GetSimpleVertexShader(),
VertexShaderCache::GetSimpleInputLayout(),
GeometryShaderCache::GetCopyGeometryShader());
D3D::context->OMSetRenderTargets(1, &FramebufferManager::GetEFBColorTexture()->GetRTV(), FramebufferManager::GetEFBDepthTexture()->GetDSV());
g_renderer->RestoreAPIState();
}
void TextureCache::CopyEFB(u8* dst, u32 format, u32 native_width, u32 bytes_per_row, u32 num_blocks_y, u32 memory_stride,
PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool isIntensity, bool scaleByHalf)
{
g_encoder->Encode(dst, format, native_width, bytes_per_row, num_blocks_y, memory_stride, srcFormat, srcRect, isIntensity, scaleByHalf);
}
const char palette_shader[] =
R"HLSL(
sampler samp0 : register(s0);
Texture2DArray Tex0 : register(t0);
Buffer<uint> Tex1 : register(t1);
uniform float Multiply;
uint Convert3To8(uint v)
{
// Swizzle bits: 00000123 -> 12312312
return (v << 5) | (v << 2) | (v >> 1);
}
uint Convert4To8(uint v)
{
// Swizzle bits: 00001234 -> 12341234
return (v << 4) | v;
}
uint Convert5To8(uint v)
{
// Swizzle bits: 00012345 -> 12345123
return (v << 3) | (v >> 2);
}
uint Convert6To8(uint v)
{
// Swizzle bits: 00123456 -> 12345612
return (v << 2) | (v >> 4);
}
float4 DecodePixel_RGB5A3(uint val)
{
int r,g,b,a;
if ((val&0x8000))
{
r=Convert5To8((val>>10) & 0x1f);
g=Convert5To8((val>>5 ) & 0x1f);
b=Convert5To8((val ) & 0x1f);
a=0xFF;
}
else
{
a=Convert3To8((val>>12) & 0x7);
r=Convert4To8((val>>8 ) & 0xf);
g=Convert4To8((val>>4 ) & 0xf);
b=Convert4To8((val ) & 0xf);
}
return float4(r, g, b, a) / 255;
}
float4 DecodePixel_RGB565(uint val)
{
int r, g, b, a;
r = Convert5To8((val >> 11) & 0x1f);
g = Convert6To8((val >> 5) & 0x3f);
b = Convert5To8((val) & 0x1f);
a = 0xFF;
return float4(r, g, b, a) / 255;
}
float4 DecodePixel_IA8(uint val)
{
int i = val & 0xFF;
int a = val >> 8;
return float4(i, i, i, a) / 255;
}
void main(
out float4 ocol0 : SV_Target,
in float4 pos : SV_Position,
in float3 uv0 : TEXCOORD0)
{
uint src = round(Tex0.Sample(samp0,uv0) * Multiply).r;
src = Tex1.Load(src);
src = ((src << 8) & 0xFF00) | (src >> 8);
ocol0 = DECODE(src);
}
)HLSL";
void TextureCache::ConvertTexture(TCacheEntryBase* entry, TCacheEntryBase* unconverted, void* palette, TlutFormat format)
{
g_renderer->ResetAPIState();
// stretch picture with increased internal resolution
const D3D11_VIEWPORT vp = CD3D11_VIEWPORT(0.f, 0.f, (float)unconverted->config.width, (float)unconverted->config.height);
D3D::context->RSSetViewports(1, &vp);
D3D11_BOX box{ 0, 0, 0, 512, 1, 1 };
D3D::context->UpdateSubresource(palette_buf, 0, &box, palette, 0, 0);
D3D::stateman->SetTexture(1, palette_buf_srv);
// TODO: Add support for C14X2 format. (Different multiplier, more palette entries.)
float params[4] = { (unconverted->format & 0xf) == 0 ? 15.f : 255.f };
D3D::context->UpdateSubresource(palette_uniform, 0, nullptr, ¶ms, 0, 0);
D3D::stateman->SetPixelConstants(palette_uniform);
const D3D11_RECT sourcerect = CD3D11_RECT(0, 0, unconverted->config.width, unconverted->config.height);
D3D::SetPointCopySampler();
// Make sure we don't draw with the texture set as both a source and target.
// (This can happen because we don't unbind textures when we free them.)
D3D::stateman->UnsetTexture(static_cast<TCacheEntry*>(entry)->texture->GetSRV());
D3D::context->OMSetRenderTargets(1, &static_cast<TCacheEntry*>(entry)->texture->GetRTV(), nullptr);
// Create texture copy
D3D::drawShadedTexQuad(
static_cast<TCacheEntry*>(unconverted)->texture->GetSRV(),
&sourcerect, unconverted->config.width, unconverted->config.height,
palette_pixel_shader[format],
VertexShaderCache::GetSimpleVertexShader(), VertexShaderCache::GetSimpleInputLayout(),
GeometryShaderCache::GetCopyGeometryShader());
D3D::context->OMSetRenderTargets(1, &FramebufferManager::GetEFBColorTexture()->GetRTV(), FramebufferManager::GetEFBDepthTexture()->GetDSV());
g_renderer->RestoreAPIState();
}
ID3D11PixelShader *GetConvertShader(const char* Type)
{
std::string shader = "#define DECODE DecodePixel_";
shader.append(Type);
shader.append("\n");
shader.append(palette_shader);
return D3D::CompileAndCreatePixelShader(shader);
}
TextureCache::TextureCache()
{
// FIXME: Is it safe here?
g_encoder = std::make_unique<PSTextureEncoder>();
g_encoder->Init();
palette_buf = nullptr;
palette_buf_srv = nullptr;
palette_uniform = nullptr;
palette_pixel_shader[GX_TL_IA8] = GetConvertShader("IA8");
palette_pixel_shader[GX_TL_RGB565] = GetConvertShader("RGB565");
palette_pixel_shader[GX_TL_RGB5A3] = GetConvertShader("RGB5A3");
auto lutBd = CD3D11_BUFFER_DESC(sizeof(u16) * 256, D3D11_BIND_SHADER_RESOURCE);
HRESULT hr = D3D::device->CreateBuffer(&lutBd, nullptr, &palette_buf);
CHECK(SUCCEEDED(hr), "create palette decoder lut buffer");
D3D::SetDebugObjectName(palette_buf, "texture decoder lut buffer");
// TODO: C14X2 format.
auto outlutUavDesc = CD3D11_SHADER_RESOURCE_VIEW_DESC(palette_buf, DXGI_FORMAT_R16_UINT, 0, 256, 0);
hr = D3D::device->CreateShaderResourceView(palette_buf, &outlutUavDesc, &palette_buf_srv);
CHECK(SUCCEEDED(hr), "create palette decoder lut srv");
D3D::SetDebugObjectName(palette_buf_srv, "texture decoder lut srv");
const D3D11_BUFFER_DESC cbdesc = CD3D11_BUFFER_DESC(16, D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DEFAULT);
hr = D3D::device->CreateBuffer(&cbdesc, nullptr, &palette_uniform);
CHECK(SUCCEEDED(hr), "Create palette decoder constant buffer");
D3D::SetDebugObjectName((ID3D11DeviceChild*)palette_uniform, "a constant buffer used in TextureCache::CopyRenderTargetToTexture");
}
ConstantStreamBuffer::ConstantStreamBuffer(int size) : m_max_size(ROUND_UP(size, 256)), m_need_init(true)
{
m_use_partial_buffer_update = D3D::SupportPartialContantBufferUpdate();
D3D11_BUFFER_DESC desc = CD3D11_BUFFER_DESC(m_max_size, D3D11_BIND_CONSTANT_BUFFER, D3D11_USAGE_DYNAMIC, D3D11_CPU_ACCESS_WRITE);
device->CreateBuffer(&desc, nullptr, &m_buf);
}
