#include <dali/internal/render/renderers/render-renderer.h>
// INTERNAL INCLUDES
-#include <dali/graphics-api/graphics-program.h>
#include <dali/graphics-api/graphics-types.h>
#include <dali/integration-api/debug.h>
#include <dali/internal/common/image-sampler.h>
#include <dali/internal/render/common/render-instruction.h>
#include <dali/internal/render/data-providers/node-data-provider.h>
#include <dali/internal/render/data-providers/uniform-map-data-provider.h>
-#include <dali/internal/render/gl-resources/context.h>
#include <dali/internal/render/renderers/render-sampler.h>
#include <dali/internal/render/renderers/render-texture.h>
#include <dali/internal/render/renderers/render-vertex-buffer.h>
#include <dali/internal/render/renderers/shader-cache.h>
#include <dali/internal/render/shaders/program.h>
-#include <dali/internal/render/shaders/scene-graph-shader.h>
+#include <dali/internal/render/shaders/render-shader.h>
#include <dali/internal/update/common/uniform-map.h>
+#include <dali/internal/render/renderers/uniform-buffer-view.h>
+#include <dali/internal/render/renderers/uniform-buffer-view-pool.h>
-namespace Dali
-{
-namespace Internal
+namespace Dali::Internal
{
namespace
{
-// Size of uniform buffer page used when resizing
-constexpr uint32_t UBO_PAGE_SIZE = 8192u;
-
-// UBO allocation threshold below which the UBO will shrink
-constexpr auto UBO_SHRINK_THRESHOLD = 0.75f;
-
// Helper to get the vertex input format
Dali::Graphics::VertexInputFormat GetPropertyVertexFormat(Property::Type propertyType)
{
break;
}
case Property::MATRIX3:
- {
- type = Dali::Graphics::VertexInputFormat::FLOAT;
- break;
- }
case Property::MATRIX:
{
type = Dali::Graphics::VertexInputFormat::FLOAT;
StencilParameters& stencilParameters)
: mGraphicsController(nullptr),
mRenderDataProvider(dataProvider),
- mContext(nullptr),
mGeometry(geometry),
mProgramCache(nullptr),
mUniformIndexMap(),
mDepthWriteMode(depthWriteMode),
mDepthTestMode(depthTestMode),
mUpdateAttributeLocations(true),
- mPremultipledAlphaEnabled(preMultipliedAlphaEnabled),
+ mPremultipliedAlphaEnabled(preMultipliedAlphaEnabled),
mShaderChanged(false),
mUpdated(true)
{
mBlendingOptions.SetBlendColor(blendColor);
}
-void Renderer::Initialize(Context& context, Graphics::Controller& graphicsController, ProgramCache& programCache, Render::ShaderCache& shaderCache, Render::UniformBufferManager& uniformBufferManager)
+void Renderer::Initialize(Graphics::Controller& graphicsController, ProgramCache& programCache, Render::ShaderCache& shaderCache, Render::UniformBufferManager& uniformBufferManager)
{
- mContext = &context;
mGraphicsController = &graphicsController;
mProgramCache = &programCache;
mShaderCache = &shaderCache;
mDrawCommands.insert(mDrawCommands.end(), pDrawCommands, pDrawCommands + size);
}
-void Renderer::SetUniformFromProperty(BufferIndex bufferIndex, Program& program, UniformIndexMap& map)
-{
- GLint location = program.GetUniformLocation(map.uniformIndex);
- if(Program::UNIFORM_UNKNOWN != location)
- {
- // switch based on property type to use correct GL uniform setter
- switch(map.propertyValue->GetType())
- {
- case Property::INTEGER:
- {
- program.SetUniform1i(location, map.propertyValue->GetInteger(bufferIndex));
- break;
- }
- case Property::FLOAT:
- {
- program.SetUniform1f(location, map.propertyValue->GetFloat(bufferIndex));
- break;
- }
- case Property::VECTOR2:
- {
- Vector2 value(map.propertyValue->GetVector2(bufferIndex));
- program.SetUniform2f(location, value.x, value.y);
- break;
- }
-
- case Property::VECTOR3:
- {
- Vector3 value(map.propertyValue->GetVector3(bufferIndex));
- program.SetUniform3f(location, value.x, value.y, value.z);
- break;
- }
-
- case Property::VECTOR4:
- {
- Vector4 value(map.propertyValue->GetVector4(bufferIndex));
- program.SetUniform4f(location, value.x, value.y, value.z, value.w);
- break;
- }
-
- case Property::ROTATION:
- {
- Quaternion value(map.propertyValue->GetQuaternion(bufferIndex));
- program.SetUniform4f(location, value.mVector.x, value.mVector.y, value.mVector.z, value.mVector.w);
- break;
- }
-
- case Property::MATRIX:
- {
- const Matrix& value = map.propertyValue->GetMatrix(bufferIndex);
- program.SetUniformMatrix4fv(location, 1, value.AsFloat());
- break;
- }
-
- case Property::MATRIX3:
- {
- const Matrix3& value = map.propertyValue->GetMatrix3(bufferIndex);
- program.SetUniformMatrix3fv(location, 1, value.AsFloat());
- break;
- }
-
- default:
- {
- // Other property types are ignored
- break;
- }
- }
- }
-}
-
-void Renderer::BindTextures(Program& program, Graphics::CommandBuffer& commandBuffer, Vector<Graphics::Texture*>& boundTextures)
+void Renderer::BindTextures(Graphics::CommandBuffer& commandBuffer, Vector<Graphics::Texture*>& boundTextures)
{
uint32_t textureUnit = 0;
- GLint uniformLocation(-1);
std::vector<Render::Sampler*>& samplers(mRenderDataProvider->GetSamplers());
std::vector<Render::Texture*>& textures(mRenderDataProvider->GetTextures());
{
if(textures[i] && textures[i]->GetGraphicsObject())
{
- if(program.GetSamplerUniformLocation(i, uniformLocation))
- {
- // if the sampler exists,
- // if it's default, delete the graphics object
- // otherwise re-initialize it if dirty
+ // if the sampler exists,
+ // if it's default, delete the graphics object
+ // otherwise re-initialize it if dirty
- const Graphics::Sampler* graphicsSampler = (samplers[i] ? samplers[i]->GetGraphicsObject()
- : nullptr);
+ const Graphics::Sampler* graphicsSampler = (samplers[i] ? samplers[i]->GetGraphicsObject()
+ : nullptr);
- boundTextures.PushBack(textures[i]->GetGraphicsObject());
- const Graphics::TextureBinding textureBinding{textures[i]->GetGraphicsObject(), graphicsSampler, textureUnit};
- textureBindings.push_back(textureBinding);
+ boundTextures.PushBack(textures[i]->GetGraphicsObject());
+ const Graphics::TextureBinding textureBinding{textures[i]->GetGraphicsObject(), graphicsSampler, textureUnit};
+ textureBindings.push_back(textureBinding);
- program.SetUniform1i(uniformLocation, textureUnit); // Get through shader reflection
- ++textureUnit;
- }
+ ++textureUnit;
}
}
- if(textureBindings.size() > 0)
+ if(!textureBindings.empty())
{
commandBuffer.BindTextures(textureBindings);
}
void Renderer::EnablePreMultipliedAlpha(bool enable)
{
- mPremultipledAlphaEnabled = enable;
- mUpdated = true;
+ mPremultipliedAlphaEnabled = enable;
+ mUpdated = true;
}
void Renderer::SetDepthWriteMode(DepthWriteMode::Type depthWriteMode)
mGeometry->Upload(*mGraphicsController);
}
-bool Renderer::Render(Context& context,
+bool Renderer::Render(Graphics::CommandBuffer& commandBuffer,
BufferIndex bufferIndex,
const SceneGraph::NodeDataProvider& node,
const Matrix& modelMatrix,
return false;
}
- // Create command buffer if not present
- if(!mGraphicsCommandBuffer)
- {
- mGraphicsCommandBuffer = mGraphicsController->CreateCommandBuffer(
- Graphics::CommandBufferCreateInfo()
- .SetLevel(Graphics::CommandBufferLevel::SECONDARY),
- nullptr);
- }
- else
- {
- mGraphicsCommandBuffer->Reset();
- }
-
- auto& commandBuffer = mGraphicsCommandBuffer;
-
// Set blending mode
if(!mDrawCommands.empty())
{
- blend = (commands[0]->queue == DevelRenderer::RENDER_QUEUE_OPAQUE ? false : blend);
+ blend = (commands[0]->queue != DevelRenderer::RENDER_QUEUE_OPAQUE) && blend;
}
// Create Program
Program* program = Program::New(*mProgramCache,
shaderData,
*mGraphicsController,
- std::move(graphicsProgram),
- (shaderData->GetHints() & Dali::Shader::Hint::MODIFIES_GEOMETRY) != 0x0);
+ std::move(graphicsProgram));
if(!program)
{
return false;
}
- // Temporarily create a pipeline here - this will be used for transporting
- // topology, vertex format, attrs, rasterization state
- mGraphicsPipeline = PrepareGraphicsPipeline(*program, instruction, blend, std::move(mGraphicsPipeline));
+ // Prepare the graphics pipeline. This may either re-use an existing pipeline or create a new one.
+ auto& pipeline = PrepareGraphicsPipeline(*program, instruction, node, blend);
- commandBuffer->BindPipeline(*mGraphicsPipeline.get());
+ commandBuffer.BindPipeline(pipeline);
- BindTextures(*program, *commandBuffer.get(), boundTextures);
+ BindTextures(commandBuffer, boundTextures);
BuildUniformIndexMap(bufferIndex, node, size, *program);
- WriteUniformBuffer(bufferIndex, *commandBuffer.get(), program, instruction, node, modelMatrix, modelViewMatrix, viewMatrix, projectionMatrix, size);
+ WriteUniformBuffer(bufferIndex, commandBuffer, program, instruction, node, modelMatrix, modelViewMatrix, viewMatrix, projectionMatrix, size);
bool drawn = false; // Draw can fail if there are no vertex buffers or they haven't been uploaded yet
// @todo We should detect this case much earlier to prevent unnecessary work
- //@todo manage mDrawCommands in the same way as above command buffer?!
if(mDrawCommands.empty())
{
- drawn = mGeometry->Draw(*mGraphicsController, *commandBuffer.get(), mIndexedDrawFirstElement, mIndexedDrawElementsCount);
+ drawn = mGeometry->Draw(*mGraphicsController, commandBuffer, mIndexedDrawFirstElement, mIndexedDrawElementsCount);
}
else
{
for(auto& cmd : commands)
{
- // @todo This should generate a command buffer per cmd
- // Tests WILL fail. (Temporarily commented out)
- mGeometry->Draw(*mGraphicsController, *commandBuffer.get(), cmd->firstIndex, cmd->elementCount);
+ mGeometry->Draw(*mGraphicsController, commandBuffer, cmd->firstIndex, cmd->elementCount);
}
}
- // Command buffer contains Texture bindings, vertex bindings, index buffer binding, pipeline(vertex format)
- // @todo We should return the command buffer(s) and let the calling method submit
- // If not drawn, then don't add command buffer to submit info, and if empty, don't
- // submit.
- /*
- if(drawn)
- {
- Graphics::SubmitInfo submitInfo{{}, 0 | Graphics::SubmitFlagBits::FLUSH};
- submitInfo.cmdBuffer.push_back(commandBuffer.get());
- mGraphicsController->SubmitCommandBuffers(submitInfo);
- }
- */
mUpdated = false;
return drawn;
}
const SceneGraph::CollectedUniformMap& uniformMap = uniformMapDataProvider.GetUniformMap(bufferIndex);
const SceneGraph::CollectedUniformMap& uniformMapNode = node.GetUniformMap(bufferIndex);
- uint32_t maxMaps = static_cast<uint32_t>(uniformMap.Count() + uniformMapNode.Count()); // 4,294,967,295 maps should be enough
- mUniformIndexMap.Clear(); // Clear contents, but keep memory if we don't change size
+ auto maxMaps = static_cast<uint32_t>(uniformMap.Count() + uniformMapNode.Count()); // 4,294,967,295 maps should be enough
+ mUniformIndexMap.Clear(); // Clear contents, but keep memory if we don't change size
mUniformIndexMap.Resize(maxMaps);
+ // Copy uniform map into mUniformIndexMap
uint32_t mapIndex = 0;
for(; mapIndex < uniformMap.Count(); ++mapIndex)
{
mUniformIndexMap[mapIndex].propertyValue = uniformMap[mapIndex].propertyPtr;
- mUniformIndexMap[mapIndex].uniformIndex = program.RegisterUniform(uniformMap[mapIndex].uniformName);
mUniformIndexMap[mapIndex].uniformName = uniformMap[mapIndex].uniformName;
mUniformIndexMap[mapIndex].uniformNameHash = uniformMap[mapIndex].uniformNameHash;
mUniformIndexMap[mapIndex].uniformNameHashNoArray = uniformMap[mapIndex].uniformNameHashNoArray;
for(uint32_t nodeMapIndex = 0; nodeMapIndex < uniformMapNode.Count(); ++nodeMapIndex)
{
- uint32_t uniformIndex = program.RegisterUniform(uniformMapNode[nodeMapIndex].uniformName);
- bool found(false);
+ auto hash = uniformMapNode[nodeMapIndex].uniformNameHash;
+ auto& name = uniformMapNode[nodeMapIndex].uniformName;
+ bool found(false);
for(uint32_t i = 0; i < uniformMap.Count(); ++i)
{
- if(mUniformIndexMap[i].uniformIndex == uniformIndex)
+ if(mUniformIndexMap[i].uniformNameHash == hash &&
+ mUniformIndexMap[i].uniformName == name)
{
mUniformIndexMap[i].propertyValue = uniformMapNode[nodeMapIndex].propertyPtr;
found = true;
{
mUniformIndexMap[mapIndex].propertyValue = uniformMapNode[nodeMapIndex].propertyPtr;
mUniformIndexMap[mapIndex].uniformName = uniformMapNode[nodeMapIndex].uniformName;
- mUniformIndexMap[mapIndex].uniformIndex = uniformIndex;
mUniformIndexMap[mapIndex].uniformNameHash = uniformMapNode[nodeMapIndex].uniformNameHash;
mUniformIndexMap[mapIndex].uniformNameHashNoArray = uniformMapNode[nodeMapIndex].uniformNameHashNoArray;
mUniformIndexMap[mapIndex].arrayIndex = uniformMapNode[nodeMapIndex].arrayIndex;
uniformBlockAllocationBytes += blockSize;
}
- auto pagedAllocation = ((uniformBlockAllocationBytes / UBO_PAGE_SIZE + 1u)) * UBO_PAGE_SIZE;
-
- // Allocate twice memory as required by the uniform buffers
- // todo: memory usage backlog to use optimal allocation
- if(uniformBlockAllocationBytes && !mUniformBuffer[bufferIndex])
- {
- mUniformBuffer[bufferIndex] = mUniformBufferManager->AllocateUniformBuffer(pagedAllocation);
- }
- else if(uniformBlockAllocationBytes && (mUniformBuffer[bufferIndex]->GetSize() < pagedAllocation ||
- (pagedAllocation < uint32_t(float(mUniformBuffer[bufferIndex]->GetSize()) * UBO_SHRINK_THRESHOLD))))
+ // Create uniform buffer view from uniform buffer
+ Graphics::UniquePtr<Render::UniformBufferView> uboView {nullptr};
+ if(uniformBlockAllocationBytes)
{
- mUniformBuffer[bufferIndex]->Reserve(pagedAllocation);
- }
+ auto uboPoolView = mUniformBufferManager->GetUniformBufferViewPool( bufferIndex );
- // Clear UBO
- if(mUniformBuffer[bufferIndex])
- {
- mUniformBuffer[bufferIndex]->Fill(0, 0u, 0u);
+ uboView = uboPoolView->CreateUniformBufferView( uniformBlockAllocationBytes );
}
// update the uniform buffer
// pass shared UBO and offset, return new offset for next item to be used
// don't process bindings if there are no uniform buffers allocated
- auto ubo = mUniformBuffer[bufferIndex].get();
- if(ubo)
+ if(uboView)
{
auto uboCount = reflection.GetUniformBlockCount();
mUniformBufferBindings.resize(uboCount);
std::vector<Graphics::UniformBufferBinding>* bindings{&mUniformBufferBindings};
+ mUniformBufferBindings[0].buffer = uboView->GetBuffer( &mUniformBufferBindings[0].offset );
+
// Write default uniforms
- WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::MODEL_MATRIX), *ubo, *bindings, modelMatrix);
- WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::VIEW_MATRIX), *ubo, *bindings, viewMatrix);
- WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::PROJECTION_MATRIX), *ubo, *bindings, projectionMatrix);
- WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::MODEL_VIEW_MATRIX), *ubo, *bindings, modelViewMatrix);
+ WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::MODEL_MATRIX), *uboView, *bindings, modelMatrix);
+ WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::VIEW_MATRIX), *uboView, *bindings, viewMatrix);
+ WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::PROJECTION_MATRIX), *uboView, *bindings, projectionMatrix);
+ WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::MODEL_VIEW_MATRIX), *uboView, *bindings, modelViewMatrix);
auto mvpUniformInfo = program->GetDefaultUniform(Program::DefaultUniformIndex::MVP_MATRIX);
if(mvpUniformInfo && !mvpUniformInfo->name.empty())
{
Matrix modelViewProjectionMatrix(false);
Matrix::Multiply(modelViewProjectionMatrix, modelViewMatrix, projectionMatrix);
- WriteDefaultUniform(mvpUniformInfo, *ubo, *bindings, modelViewProjectionMatrix);
+ WriteDefaultUniform(mvpUniformInfo, *uboView, *bindings, modelViewProjectionMatrix);
}
auto normalUniformInfo = program->GetDefaultUniform(Program::DefaultUniformIndex::NORMAL_MATRIX);
Matrix3 normalMatrix(modelViewMatrix);
normalMatrix.Invert();
normalMatrix.Transpose();
- WriteDefaultUniform(normalUniformInfo, *ubo, *bindings, normalMatrix);
+ WriteDefaultUniform(normalUniformInfo, *uboView, *bindings, normalMatrix);
}
Vector4 finalColor;
const Vector4& color = node.GetRenderColor(bufferIndex);
- if(mPremultipledAlphaEnabled)
+ if(mPremultipliedAlphaEnabled)
{
float alpha = color.a * mRenderDataProvider->GetOpacity(bufferIndex);
finalColor = Vector4(color.r * alpha, color.g * alpha, color.b * alpha, alpha);
{
finalColor = Vector4(color.r, color.g, color.b, color.a * mRenderDataProvider->GetOpacity(bufferIndex));
}
- WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::COLOR), *ubo, *bindings, finalColor);
+ WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::COLOR), *uboView, *bindings, finalColor);
// Write uniforms from the uniform map
- FillUniformBuffer(*program, instruction, *ubo, bindings, uboOffset, bufferIndex);
+ FillUniformBuffer(*program, instruction, *uboView, bindings, uboOffset, bufferIndex);
// Write uSize in the end, as it shouldn't be overridable by dynamic properties.
- WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::SIZE), *ubo, *bindings, size);
+ WriteDefaultUniform(program->GetDefaultUniform(Program::DefaultUniformIndex::SIZE), *uboView, *bindings, size);
commandBuffer.BindUniformBuffers(*bindings);
}
}
template<class T>
-bool Renderer::WriteDefaultUniform(const Graphics::UniformInfo* uniformInfo, Render::UniformBuffer& ubo, const std::vector<Graphics::UniformBufferBinding>& bindings, const T& data)
+bool Renderer::WriteDefaultUniform(const Graphics::UniformInfo* uniformInfo, Render::UniformBufferView& ubo, const std::vector<Graphics::UniformBufferBinding>& bindings, const T& data)
{
if(uniformInfo && !uniformInfo->name.empty())
{
}
template<class T>
-void Renderer::WriteUniform(Render::UniformBuffer& ubo, const std::vector<Graphics::UniformBufferBinding>& bindings, const Graphics::UniformInfo& uniformInfo, const T& data)
+void Renderer::WriteUniform(Render::UniformBufferView& ubo, const std::vector<Graphics::UniformBufferBinding>& bindings, const Graphics::UniformInfo& uniformInfo, const T& data)
{
WriteUniform(ubo, bindings, uniformInfo, &data, sizeof(T));
}
-void Renderer::WriteUniform(Render::UniformBuffer& ubo, const std::vector<Graphics::UniformBufferBinding>& bindings, const Graphics::UniformInfo& uniformInfo, const void* data, uint32_t size)
+void Renderer::WriteUniform(Render::UniformBufferView& ubo, const std::vector<Graphics::UniformBufferBinding>& bindings, const Graphics::UniformInfo& uniformInfo, const void* data, uint32_t size)
{
- ubo.Write(data, size, bindings[uniformInfo.bufferIndex].offset + uniformInfo.offset);
+ ubo.Write(data, size, ubo.GetOffset() + uniformInfo.offset);
}
void Renderer::FillUniformBuffer(Program& program,
const SceneGraph::RenderInstruction& instruction,
- Render::UniformBuffer& ubo,
+ Render::UniformBufferView& ubo,
std::vector<Graphics::UniformBufferBinding>*& outBindings,
uint32_t& offset,
BufferIndex updateBufferIndex)
{
mUniformBufferBindings[i].dataSize = reflection.GetUniformBlockSize(i);
mUniformBufferBindings[i].binding = reflection.GetUniformBlockBinding(i);
- mUniformBufferBindings[i].offset = dataOffset;
dataOffset += GetUniformBufferDataAlignment(mUniformBufferBindings[i].dataSize);
- mUniformBufferBindings[i].buffer = ubo.GetBuffer();
+ mUniformBufferBindings[i].buffer = ubo.GetBuffer( &mUniformBufferBindings[i].offset );
for(UniformIndexMappings::Iterator iter = mUniformIndexMap.Begin(),
end = mUniformIndexMap.End();
if(uniformFound)
{
- auto dst = mUniformBufferBindings[uniformInfo.bufferIndex].offset + uniformInfo.offset;
-
+ auto dst = ubo.GetOffset() + uniformInfo.offset;
switch((*iter).propertyValue->GetType())
{
case Property::Type::BOOLEAN:
offset = dataOffset;
}
-void Renderer::SetSortAttributes(BufferIndex bufferIndex,
- SceneGraph::RenderInstructionProcessor::SortAttributes& sortAttributes) const
+void Renderer::SetSortAttributes(SceneGraph::RenderInstructionProcessor::SortAttributes& sortAttributes) const
{
sortAttributes.shader = &(mRenderDataProvider->GetShader());
sortAttributes.geometry = mGeometry;
return false;
}
-Graphics::UniquePtr<Graphics::Pipeline> Renderer::PrepareGraphicsPipeline(
+Graphics::Pipeline& Renderer::PrepareGraphicsPipeline(
Program& program,
const Dali::Internal::SceneGraph::RenderInstruction& instruction,
- bool blend,
- Graphics::UniquePtr<Graphics::Pipeline>&& oldPipeline)
+ const SceneGraph::NodeDataProvider& node,
+ bool blend)
{
Graphics::InputAssemblyState inputAssemblyState{};
Graphics::VertexInputState vertexInputState{};
mAttributeLocations.PushBack(pLocation);
}
- uint32_t location = static_cast<uint32_t>(mAttributeLocations[base + i]);
+ auto location = static_cast<uint32_t>(mAttributeLocations[base + i]);
vertexInputState.attributes.emplace_back(location,
bindingIndex,
break;
}
- // @todo How to signal a blend barrier is needed?
- //if(mBlendingOptions.IsAdvancedBlendEquationApplied() && mPremultipledAlphaEnabled)
- //{
- // context.BlendBarrier();
- //}
+ // @todo Add blend barrier to the Graphics API if we are using advanced
+ // blending options. Command?
Graphics::ColorBlendState colorBlendState{};
colorBlendState.SetBlendEnable(false);
Graphics::BlendOp rgbOp = ConvertBlendEquation(mBlendingOptions.GetBlendEquationRgb());
Graphics::BlendOp alphaOp = ConvertBlendEquation(mBlendingOptions.GetBlendEquationRgb());
- if(mBlendingOptions.IsAdvancedBlendEquationApplied() && mPremultipledAlphaEnabled)
+ if(mBlendingOptions.IsAdvancedBlendEquationApplied() && mPremultipliedAlphaEnabled)
{
if(rgbOp != alphaOp)
{
.SetAlphaBlendOp(alphaOp);
// Blend color is optional and rarely used
- Vector4* blendColor = const_cast<Vector4*>(mBlendingOptions.GetBlendColor());
+ auto* blendColor = const_cast<Vector4*>(mBlendingOptions.GetBlendColor());
if(blendColor)
{
colorBlendState.SetBlendConstants(blendColor->AsFloat());
}
}
- // Take the program into use so we can send uniforms to it
- // @todo Remove this call entirely!
- program.Use();
-
mUpdated = true;
- // Create a new pipeline
- // @todo Passed as pointers - shallow copy will break. Implementation MUST deep copy.
- return mGraphicsController->CreatePipeline(
- Graphics::PipelineCreateInfo()
- .SetInputAssemblyState(&inputAssemblyState)
- .SetVertexInputState(&vertexInputState)
- .SetRasterizationState(&rasterizationState)
- .SetColorBlendState(&colorBlendState)
- .SetProgramState(&programState)
- .SetNextExtension(&mLegacyProgram),
- std::move(oldPipeline));
+ // Create the pipeline
+ Graphics::PipelineCreateInfo createInfo;
+ createInfo
+ .SetInputAssemblyState(&inputAssemblyState)
+ .SetVertexInputState(&vertexInputState)
+ .SetRasterizationState(&rasterizationState)
+ .SetColorBlendState(&colorBlendState)
+ .SetProgramState(&programState);
+
+ // Store a pipeline per renderer per render (renderer can be owned by multiple nodes,
+ // and re-drawn in multiple instructions).
+ // @todo This is only needed because ColorBlend state can change. Fixme!
+ // This is ameliorated by the fact that implementation caches pipelines, and we're only storing
+ // handles.
+ auto hash = HashedPipeline::GetHash(&node, &instruction, blend);
+ HashedPipeline* hashedPipeline = nullptr;
+ for(auto& element : mGraphicsPipelines)
+ {
+ if(element.mHash == hash)
+ {
+ hashedPipeline = &element;
+ break;
+ }
+ }
+
+ if(hashedPipeline != nullptr)
+ {
+ hashedPipeline->mGraphicsPipeline = mGraphicsController->CreatePipeline(
+ createInfo,
+ std::move(hashedPipeline->mGraphicsPipeline));
+ }
+ else
+ {
+ mGraphicsPipelines.emplace_back();
+ mGraphicsPipelines.back().mHash = hash;
+ mGraphicsPipelines.back().mGraphicsPipeline = mGraphicsController->CreatePipeline(createInfo, nullptr);
+ hashedPipeline = &mGraphicsPipelines.back();
+ }
+ return *hashedPipeline->mGraphicsPipeline.get();
}
} // namespace Render
-} // namespace Internal
-
} // namespace Dali
projects / platform / core / uifw / dali-core.git / blobdiff