BlueRose bd415a90d1 vault backup: 2024-02-06 10:30:09

2024-02-06 10:30:09 +08:00

36 KiB

Raw Blame History

title, date, excerpt, tags, rating

title	date	excerpt	tags	rating
Untitled	2024-02-04 12:57:56			⭐

前言

原文地址:https://www.cnblogs.com/timlly/p/15156626.html

概念

FRenderResource：是渲染线程的渲染资源基础父类，实际的数据和逻辑由子类实现。可以认为是渲染线程中承载CPU相关相关渲染的载体。
- 比如输入的顶点数据、顶点Index数据、贴图数据等。
FRHIResource：抽象了GPU侧的资源，也是众多RHI资源类型的父类。可以认为是承载显卡API相关资源的载体。
- 比如TextureSampler、TextureObject等。
FRHICommand：其父类为FRHICommandBase结构体。其含有FRHICommandBase* Next用来保存下一个Command的指针，所以存储他的结构为链表。
- 含有接口：void void ExecuteAndDestruct(FRHICommandListBase& CmdList, FRHICommandListDebugContext& DebugContext)。执行完就销毁。
- UE使用FRHICOMMAND_MACRO宏来快速定义各种RHICommand。主要功能包含：
  - 数据和资源的设置、更新、清理、转换、拷贝、回读。
  - 图元绘制。
  - Pass、SubPass、场景、ViewPort等的开始和结束事件。
  - 栅栏、等待、广播接口。
  - 光线追踪。
  - Slate、调试相关的命令。
FRHICommandList：是RHI的指令队列，用来管理、执行一组FRHICommand的对象。
- 其子类有FRHICommandListImmediate（立即执行队列）、FRHIComputeCommandList_RecursiveHazardous与TRHIComputeCommandList_RecursiveHazardous（命令列表的递归使用）
IRHICommandContext：是RHI的命令上下文接口类，定义了一组图形API相关的操作。在可以并行处理命令列表的平台上，它是一个单独的对象类。
- 主要的接口函数有：
  - 派发ComputeShader
  - 渲染查询（可见性？）
  - 相关开始/结束函数。
  - 设置数据（Viewport、GraphicsPipelineState等）
  - 设置ShadserParameter
  - 绘制图元
  - 纹理拷贝/更新
  - Raytracing
- IRHICommandContext的接口和FRHICommandList的接口高度相似且重叠。IRHICommandContext还有许多子类：
- IRHICommandContextPSOFallback：不支持真正的图形管道的RHI命令上下文。
  - FNullDynamicRHI：空实现的动态绑定RHI。
  - FOpenGLDynamicRHI：OpenGL的动态RHI。
  - FD3D11DynamicRHI：D3D11的动态RHI。
- FMetalRHICommandContext：Metal平台的命令上下文。
- FD3D12CommandContextBase：D3D12的命令上下文。
- FVulkanCommandListContext：Vulkan平台的命令队列上下文。
- FEmptyDynamicRHI：动态绑定的RHI实现的接口。
- FValidationContext：校验上下文。
IRHICommandContextContainer：IRHICommandContextContainer就是包含了IRHICommandContext对象的类型。相当于存储了一个或一组命令上下文的容器，以支持并行化地提交命令队列，只在D3D12、Metal、Vulkan等现代图形API中有实现。
- D3D12存储了FD3D12Adapter* Adapter、FD3D12CommandContext* CmdContext、 FD3D12CommandContextRedirector* CmdContextRedirector。
FDynamicRHI：FDynamicRHI是由动态绑定的RHI实现的接口，它定义的接口和CommandList、CommandContext比较相似。
- 代码详见#FDynamicRHI
FRHICommandListExecutor：负责将Renderer层的RHI中间指令转译（或直接调用）到目标平台的图形API，它在RHI体系中起着举足轻重的作用。
FParallelCommandListSet：用于实现并行渲染。使用案例详见#FParallelCommandListSet。目前5.3只有下面2个子类：
- FRDGParallelCommandListSet
- FShadowParallelCommandListSet

FDynamicRHI

class RHI_API FDynamicRHI
{
public:
    virtual ~FDynamicRHI() {}

    virtual void Init() = 0;
    virtual void PostInit() {}
    virtual void Shutdown() = 0;

    void InitPixelFormatInfo(const TArray<uint32>& PixelFormatBlockBytesIn);

    // ---- RHI接口 ----

    // 下列接口要求FlushType: Thread safe
    virtual FSamplerStateRHIRef RHICreateSamplerState(const FSamplerStateInitializerRHI& Initializer) = 0;
    virtual FRasterizerStateRHIRef RHICreateRasterizerState(const FRasterizerStateInitializerRHI& Initializer) = 0;
    virtual FDepthStencilStateRHIRef RHICreateDepthStencilState(const FDepthStencilStateInitializerRHI& Initializer) = 0;
    virtual FBlendStateRHIRef RHICreateBlendState(const FBlendStateInitializerRHI& Initializer) = 0;

    // 下列接口要求FlushType: Wait RHI Thread
    virtual FVertexDeclarationRHIRef RHICreateVertexDeclaration(const FVertexDeclarationElementList& Elements) = 0;
    virtual FPixelShaderRHIRef RHICreatePixelShader(TArrayView<const uint8> Code, const FSHAHash& Hash) = 0;
    virtual FVertexShaderRHIRef RHICreateVertexShader(TArrayView<const uint8> Code, const FSHAHash& Hash) = 0;
    virtual FHullShaderRHIRef RHICreateHullShader(TArrayView<const uint8> Code, const FSHAHash& Hash) = 0;
    virtual FDomainShaderRHIRef RHICreateDomainShader(TArrayView<const uint8> Code, const FSHAHash& Hash) = 0;
    virtual FGeometryShaderRHIRef RHICreateGeometryShader(TArrayView<const uint8> Code, const FSHAHash& Hash) = 0;
    virtual FComputeShaderRHIRef RHICreateComputeShader(TArrayView<const uint8> Code, const FSHAHash& Hash) = 0;

     // FlushType: Must be Thread-Safe.
    virtual FRenderQueryPoolRHIRef RHICreateRenderQueryPool(ERenderQueryType QueryType, uint32 NumQueries = UINT32_MAX);
    inline FComputeFenceRHIRef RHICreateComputeFence(const FName& Name);
    
    virtual FGPUFenceRHIRef RHICreateGPUFence(const FName &Name);
    virtual void RHICreateTransition(FRHITransition* Transition, ERHIPipeline SrcPipelines, ERHIPipeline DstPipelines, ERHICreateTransitionFlags CreateFlags, TArrayView<const FRHITransitionInfo> Infos);
    virtual void RHIReleaseTransition(FRHITransition* Transition);

    // FlushType: Thread safe.    
    virtual FStagingBufferRHIRef RHICreateStagingBuffer();
    virtual void* RHILockStagingBuffer(FRHIStagingBuffer* StagingBuffer, FRHIGPUFence* Fence, uint32 Offset, uint32 SizeRHI);
    virtual void RHIUnlockStagingBuffer(FRHIStagingBuffer* StagingBuffer);
    
    // FlushType: Thread safe, but varies depending on the RHI
    virtual FBoundShaderStateRHIRef RHICreateBoundShaderState(FRHIVertexDeclaration* VertexDeclaration, FRHIVertexShader* VertexShader, FRHIHullShader* HullShader, FRHIDomainShader* DomainShader, FRHIPixelShader* PixelShader, FRHIGeometryShader* GeometryShader) = 0;
    // FlushType: Thread safe
    virtual FGraphicsPipelineStateRHIRef RHICreateGraphicsPipelineState(const FGraphicsPipelineStateInitializer& Initializer);
    
    // FlushType: Thread safe, but varies depending on the RHI
    virtual FUniformBufferRHIRef RHICreateUniformBuffer(const void* Contents, const FRHIUniformBufferLayout& Layout, EUniformBufferUsage Usage, EUniformBufferValidation Validation) = 0;
    virtual void RHIUpdateUniformBuffer(FRHIUniformBuffer* UniformBufferRHI, const void* Contents) = 0;
    
    // FlushType: Wait RHI Thread
    virtual FIndexBufferRHIRef RHICreateIndexBuffer(uint32 Stride, uint32 Size, uint32 InUsage, ERHIAccess InResourceState, FRHIResourceCreateInfo& CreateInfo) = 0;
    virtual void* RHILockIndexBuffer(FRHICommandListImmediate& RHICmdList, FRHIIndexBuffer* IndexBuffer, uint32 Offset, uint32 Size, EResourceLockMode LockMode);
    virtual void RHIUnlockIndexBuffer(FRHICommandListImmediate& RHICmdList, FRHIIndexBuffer* IndexBuffer);
    virtual void RHITransferIndexBufferUnderlyingResource(FRHIIndexBuffer* DestIndexBuffer, FRHIIndexBuffer* SrcIndexBuffer);

    // FlushType: Wait RHI Thread
    virtual FVertexBufferRHIRef RHICreateVertexBuffer(uint32 Size, uint32 InUsage, ERHIAccess InResourceState, FRHIResourceCreateInfo& CreateInfo) = 0;
    // FlushType: Flush RHI Thread
    virtual void* RHILockVertexBuffer(FRHICommandListImmediate& RHICmdList, FRHIVertexBuffer* VertexBuffer, uint32 Offset, uint32 SizeRHI, EResourceLockMode LockMode);
    virtual void RHIUnlockVertexBuffer(FRHICommandListImmediate& RHICmdList, FRHIVertexBuffer* VertexBuffer);
    // FlushType: Flush Immediate (seems dangerous)
    virtual void RHICopyVertexBuffer(FRHIVertexBuffer* SourceBuffer, FRHIVertexBuffer* DestBuffer) = 0;
    virtual void RHITransferVertexBufferUnderlyingResource(FRHIVertexBuffer* DestVertexBuffer, FRHIVertexBuffer* SrcVertexBuffer);

    // FlushType: Wait RHI Thread
    virtual FStructuredBufferRHIRef RHICreateStructuredBuffer(uint32 Stride, uint32 Size, uint32 InUsage, ERHIAccess InResourceState, FRHIResourceCreateInfo& CreateInfo) = 0;
    // FlushType: Flush RHI Thread
    virtual void* RHILockStructuredBuffer(FRHICommandListImmediate& RHICmdList, FRHIStructuredBuffer* StructuredBuffer, uint32 Offset, uint32 SizeRHI, EResourceLockMode LockMode);
    virtual void RHIUnlockStructuredBuffer(FRHICommandListImmediate& RHICmdList, FRHIStructuredBuffer* StructuredBuffer);

    // FlushType: Wait RHI Thread
    virtual FUnorderedAccessViewRHIRef RHICreateUnorderedAccessView(FRHIStructuredBuffer* StructuredBuffer, bool bUseUAVCounter, bool bAppendBuffer) = 0;
    // FlushType: Wait RHI Thread
    virtual FUnorderedAccessViewRHIRef RHICreateUnorderedAccessView(FRHITexture* Texture, uint32 MipLevel) = 0;
    // FlushType: Wait RHI Thread
    virtual FUnorderedAccessViewRHIRef RHICreateUnorderedAccessView(FRHITexture* Texture, uint32 MipLevel, uint8 Format);

    (......)

    // RHI帧更新，须从主线程调用，FlushType: Thread safe
    virtual void RHITick(float DeltaTime) = 0;
    // 阻塞CPU直到GPU执行完成变成空闲. FlushType: Flush Immediate (seems wrong)
    virtual void RHIBlockUntilGPUIdle() = 0;
    // 开始当前帧，并确保GPU正在积极地工作 FlushType: Flush Immediate (copied from RHIBlockUntilGPUIdle)
    virtual void RHISubmitCommandsAndFlushGPU() {};

    // 通知RHI准备暂停它.
    virtual void RHIBeginSuspendRendering() {};
    // 暂停RHI渲染并将控制权交给系统的操作, FlushType: Thread safe
    virtual void RHISuspendRendering() {};
    // 继续RHI渲染, FlushType: Thread safe
    virtual void RHIResumeRendering() {};
    // FlushType: Flush Immediate
    virtual bool RHIIsRenderingSuspended() { return false; };

    // FlushType: called from render thread when RHI thread is flushed 
    // 仅在FRHIResource::FlushPendingDeletes内的延迟删除之前每帧调用.
    virtual void RHIPerFrameRHIFlushComplete();

    // 执行命令队列, FlushType: Wait RHI Thread
    virtual void RHIExecuteCommandList(FRHICommandList* CmdList) = 0;

    // FlushType: Flush RHI Thread
    virtual void* RHIGetNativeDevice() = 0;
    // FlushType: Flush RHI Thread
    virtual void* RHIGetNativeInstance() = 0;

    // 获取命令上下文. FlushType: Thread safe
    virtual IRHICommandContext* RHIGetDefaultContext() = 0;
    // 获取计算上下文. FlushType: Thread safe
    virtual IRHIComputeContext* RHIGetDefaultAsyncComputeContext();

    // FlushType: Thread safe
    virtual class IRHICommandContextContainer* RHIGetCommandContextContainer(int32 Index, int32 Num) = 0;

    // 直接由渲染线程调用的接口, 以优化RHI调用.
    virtual FVertexBufferRHIRef CreateAndLockVertexBuffer_RenderThread(class FRHICommandListImmediate& RHICmdList, uint32 Size, uint32 InUsage, ERHIAccess InResourceState, FRHIResourceCreateInfo& CreateInfo, void*& OutDataBuffer);
    virtual FIndexBufferRHIRef CreateAndLockIndexBuffer_RenderThread(class FRHICommandListImmediate& RHICmdList, uint32 Stride, uint32 Size, uint32 InUsage, ERHIAccess InResourceState, FRHIResourceCreateInfo& CreateInfo, void*& OutDataBuffer);
    
    (......)

    // Buffer Lock/Unlock
    virtual void* LockVertexBuffer_BottomOfPipe(class FRHICommandListImmediate& RHICmdList, ...);
    virtual void* LockIndexBuffer_BottomOfPipe(class FRHICommandListImmediate& RHICmdList, ...);
    
    (......)
};

以上只显示了部分接口，其中部分接口要求从渲染线程调用，部分须从游戏线程调用。大多数接口在被调用前需刷新指定类型的命令，比如：

class RHI_API FDynamicRHI
{
    // FlushType: Wait RHI Thread
    void RHIExecuteCommandList(FRHICommandList* CmdList);

    // FlushType: Flush Immediate
    void RHIBlockUntilGPUIdle();

    // FlushType: Thread safe 
    void RHITick(float DeltaTime);
};

可以在FRHICommandListImmediate的ExecuteCommandList()、BlockUntilGPUIdle()、Tick() 看到调用。

需要注意的是，传统图形API（D3D11、OpenGL）除了继承FDynamicRHI，还需要继承IRHICommandContextPSOFallback，因为需要借助后者的接口处理PSO的数据和行为，以保证传统和现代API对PSO的一致处理行为。也正因为此，现代图形API（D3D12、Vulkan、Metal）不需要继承IRHICommandContext的任何继承体系的类型，单单直接继承FDynamicRHI就可以处理RHI层的所有数据和操作。既然现代图形API（D3D12、Vulkan、Metal）的DynamicRHI没有继承IRHICommandContext的任何继承体系的类型，那么它们是如何实现FDynamicRHI::RHIGetDefaultContext的接口？下面以FD3D12DynamicRHI为例：

FParallelCommandListSet

//Engine\Source\Runtime\Renderer\Private\DepthRendering.cpp
void FDeferredShadingSceneRenderer::RenderPrePass(FRDGBuilder& GraphBuilder, FRDGTextureRef SceneDepthTexture, FInstanceCullingManager& InstanceCullingManager, FRDGTextureRef* FirstStageDepthBuffer)
{
	RDG_EVENT_SCOPE(GraphBuilder, "PrePass %s %s", GetDepthDrawingModeString(DepthPass.EarlyZPassMode), GetDepthPassReason(DepthPass.bDitheredLODTransitionsUseStencil, ShaderPlatform));
	RDG_CSV_STAT_EXCLUSIVE_SCOPE(GraphBuilder, RenderPrePass);
	RDG_GPU_STAT_SCOPE(GraphBuilder, Prepass);

	SCOPED_NAMED_EVENT(FDeferredShadingSceneRenderer_RenderPrePass, FColor::Emerald);
	SCOPE_CYCLE_COUNTER(STAT_DepthDrawTime);

	const bool bParallelDepthPass = GRHICommandList.UseParallelAlgorithms() && CVarParallelPrePass.GetValueOnRenderThread();

	RenderPrePassHMD(GraphBuilder, SceneDepthTexture);

	if (DepthPass.IsRasterStencilDitherEnabled())
	{
		AddDitheredStencilFillPass(GraphBuilder, Views, SceneDepthTexture, DepthPass);
	}

	auto RenderDepthPass = [&](uint8 DepthMeshPass)
	{
		check(DepthMeshPass == EMeshPass::DepthPass || DepthMeshPass == EMeshPass::SecondStageDepthPass);
		const bool bSecondStageDepthPass = DepthMeshPass == EMeshPass::SecondStageDepthPass;

		if (bParallelDepthPass)
		{
			RDG_WAIT_FOR_TASKS_CONDITIONAL(GraphBuilder, IsDepthPassWaitForTasksEnabled());

			for (int32 ViewIndex = 0; ViewIndex < Views.Num(); ++ViewIndex)
			{
				FViewInfo& View = Views[ViewIndex];
				RDG_GPU_MASK_SCOPE(GraphBuilder, View.GPUMask);
				RDG_EVENT_SCOPE_CONDITIONAL(GraphBuilder, Views.Num() > 1, "View%d", ViewIndex);

				FMeshPassProcessorRenderState DrawRenderState;
				SetupDepthPassState(DrawRenderState);

				const bool bShouldRenderView = View.ShouldRenderView() && (bSecondStageDepthPass ? View.bUsesSecondStageDepthPass : true);
				if (bShouldRenderView)
				{
					View.BeginRenderView();

					FDepthPassParameters* PassParameters = GetDepthPassParameters(GraphBuilder, View, SceneDepthTexture);
					View.ParallelMeshDrawCommandPasses[DepthMeshPass].BuildRenderingCommands(GraphBuilder, Scene->GPUScene, PassParameters->InstanceCullingDrawParams);

					GraphBuilder.AddPass(
						bSecondStageDepthPass ? RDG_EVENT_NAME("SecondStageDepthPassParallel") : RDG_EVENT_NAME("DepthPassParallel"),
						PassParameters,
						ERDGPassFlags::Raster | ERDGPassFlags::SkipRenderPass,
						[this, &View, PassParameters, DepthMeshPass](const FRDGPass* InPass, FRHICommandListImmediate& RHICmdList)
					{
						//并行渲染逻辑主要在这里
						FRDGParallelCommandListSet ParallelCommandListSet(InPass, RHICmdList, GET_STATID(STAT_CLP_Prepass), View, FParallelCommandListBindings(PassParameters));
						ParallelCommandListSet.SetHighPriority();
						View.ParallelMeshDrawCommandPasses[DepthMeshPass].DispatchDraw(&ParallelCommandListSet, RHICmdList, &PassParameters->InstanceCullingDrawParams);
					});

					RenderPrePassEditorPrimitives(GraphBuilder, View, PassParameters, DrawRenderState, DepthPass.EarlyZPassMode, InstanceCullingManager);
				}
			}
		}
···	
}

//Engine\Source\Runtime\Renderer\Private\MeshDrawCommands.cpp
void FParallelMeshDrawCommandPass::DispatchDraw(FParallelCommandListSet* ParallelCommandListSet, FRHICommandList& RHICmdList, const FInstanceCullingDrawParams* InstanceCullingDrawParams) const
{
	TRACE_CPUPROFILER_EVENT_SCOPE(ParallelMdcDispatchDraw);
	if (MaxNumDraws <= 0)
	{
		return;
	}

	FMeshDrawCommandOverrideArgs OverrideArgs; 
	if (InstanceCullingDrawParams)
	{
		OverrideArgs = GetMeshDrawCommandOverrideArgs(*InstanceCullingDrawParams);
	}

	if (ParallelCommandListSet)
	{
		const ENamedThreads::Type RenderThread = ENamedThreads::GetRenderThread();

		FGraphEventArray Prereqs;
		if (ParallelCommandListSet->GetPrereqs())
		{
			Prereqs.Append(*ParallelCommandListSet->GetPrereqs());
		}
		if (TaskEventRef.IsValid())
		{
			Prereqs.Add(TaskEventRef);
		}

		// Distribute work evenly to the available task graph workers based on NumEstimatedDraws.
		// Every task will then adjust it's working range based on FVisibleMeshDrawCommandProcessTask results.
		const int32 NumThreads = FMath::Min<int32>(FTaskGraphInterface::Get().GetNumWorkerThreads(), ParallelCommandListSet->Width);
		const int32 NumTasks = FMath::Min<int32>(NumThreads, FMath::DivideAndRoundUp(MaxNumDraws, ParallelCommandListSet->MinDrawsPerCommandList));
		const int32 NumDrawsPerTask = FMath::DivideAndRoundUp(MaxNumDraws, NumTasks);

		for (int32 TaskIndex = 0; TaskIndex < NumTasks; TaskIndex++)
		{
			const int32 StartIndex = TaskIndex * NumDrawsPerTask;
			const int32 NumDraws = FMath::Min(NumDrawsPerTask, MaxNumDraws - StartIndex);
			checkSlow(NumDraws > 0);

			FRHICommandList* CmdList = ParallelCommandListSet->NewParallelCommandList();

			FGraphEventRef AnyThreadCompletionEvent = TGraphTask<FDrawVisibleMeshCommandsAnyThreadTask>::CreateTask(&Prereqs, RenderThread)
				.ConstructAndDispatchWhenReady(*CmdList, TaskContext.InstanceCullingContext, TaskContext.MeshDrawCommands, TaskContext.MinimalPipelineStatePassSet,
					OverrideArgs,
					TaskContext.InstanceFactor,
					TaskIndex, NumTasks);

			ParallelCommandListSet->AddParallelCommandList(CmdList, AnyThreadCompletionEvent, NumDraws);
		}
	}
	else
	{
		QUICK_SCOPE_CYCLE_COUNTER(STAT_MeshPassDrawImmediate);

		WaitForMeshPassSetupTask(IsInActualRenderingThread() ? EWaitThread::Render : EWaitThread::Task);

		if (TaskContext.bUseGPUScene)
		{
			if (TaskContext.MeshDrawCommands.Num() > 0)
			{
				TaskContext.InstanceCullingContext.SubmitDrawCommands(
					TaskContext.MeshDrawCommands,
					TaskContext.MinimalPipelineStatePassSet,
					OverrideArgs,
					0,
					TaskContext.MeshDrawCommands.Num(),
					TaskContext.InstanceFactor,
					RHICmdList);
			}
		}
		else
		{
			SubmitMeshDrawCommandsRange(TaskContext.MeshDrawCommands, TaskContext.MinimalPipelineStatePassSet, nullptr, 0, 0, TaskContext.bDynamicInstancing, 0, TaskContext.MeshDrawCommands.Num(), TaskContext.InstanceFactor, RHICmdList);
		}
	}
}

普通Pass渲染

// 代码为UE5旧版本代码
// Engine\Source\Runtime\RHI\Public\RHIResources.h

// 渲染通道信息.
struct FRHIRenderPassInfo
{
    // 渲染纹理信息.
    struct FColorEntry
    {
        FRHITexture* RenderTarget;
        FRHITexture* ResolveTarget;
        int32 ArraySlice;
        uint8 MipIndex;
        ERenderTargetActions Action;
    };
    FColorEntry ColorRenderTargets[MaxSimultaneousRenderTargets];

    // 深度模板信息.
    struct FDepthStencilEntry
    {
        FRHITexture* DepthStencilTarget;
        FRHITexture* ResolveTarget;
        EDepthStencilTargetActions Action;
        FExclusiveDepthStencil ExclusiveDepthStencil;
    };
    FDepthStencilEntry DepthStencilRenderTarget;

    // 解析参数.
    FResolveParams ResolveParameters;

    // 部分RHI可以使用纹理来控制不同区域的采样和/或阴影分辨率
    FTextureRHIRef FoveationTexture = nullptr;

    // 部分RHI需要一个提示，遮挡查询将在这个渲染通道中使用
    uint32 NumOcclusionQueries = 0;
    bool bOcclusionQueries = false;

    // 部分RHI需要知道，在为部分资源转换生成mip映射的情况下，这个渲染通道是否将读取和写入相同的纹理.
    bool bGeneratingMips = false;

    // 如果这个renderpass应该是多视图，则需要多少视图.
    uint8 MultiViewCount = 0;

    // 部分RHI的提示，渲染通道将有特定的子通道.
    ESubpassHint SubpassHint = ESubpassHint::None;

    // 是否太多UAV.
    bool bTooManyUAVs = false;
    bool bIsMSAA = false;

    // 不同的构造函数.
    
    // Color, no depth, optional resolve, optional mip, optional array slice
    explicit FRHIRenderPassInfo(FRHITexture* ColorRT, ERenderTargetActions ColorAction, FRHITexture* ResolveRT = nullptr, uint32 InMipIndex = 0, int32 InArraySlice = -1);
    // Color MRTs, no depth
    explicit FRHIRenderPassInfo(int32 NumColorRTs, FRHITexture* ColorRTs[], ERenderTargetActions ColorAction);
    // Color MRTs, no depth
    explicit FRHIRenderPassInfo(int32 NumColorRTs, FRHITexture* ColorRTs[], ERenderTargetActions ColorAction, FRHITexture* ResolveTargets[]);
    // Color MRTs and depth
    explicit FRHIRenderPassInfo(int32 NumColorRTs, FRHITexture* ColorRTs[], ERenderTargetActions ColorAction, FRHITexture* DepthRT, EDepthStencilTargetActions DepthActions, FExclusiveDepthStencil InEDS = FExclusiveDepthStencil::DepthWrite_StencilWrite);
    // Color MRTs and depth
    explicit FRHIRenderPassInfo(int32 NumColorRTs, FRHITexture* ColorRTs[], ERenderTargetActions ColorAction, FRHITexture* ResolveRTs[], FRHITexture* DepthRT, EDepthStencilTargetActions DepthActions, FRHITexture* ResolveDepthRT, FExclusiveDepthStencil InEDS = FExclusiveDepthStencil::DepthWrite_StencilWrite);
    // Depth, no color
    explicit FRHIRenderPassInfo(FRHITexture* DepthRT, EDepthStencilTargetActions DepthActions, FRHITexture* ResolveDepthRT = nullptr, FExclusiveDepthStencil InEDS = FExclusiveDepthStencil::DepthWrite_StencilWrite);
    // Depth, no color, occlusion queries
    explicit FRHIRenderPassInfo(FRHITexture* DepthRT, uint32 InNumOcclusionQueries, EDepthStencilTargetActions DepthActions, FRHITexture* ResolveDepthRT = nullptr, FExclusiveDepthStencil InEDS = FExclusiveDepthStencil::DepthWrite_StencilWrite);
    // Color and depth
    explicit FRHIRenderPassInfo(FRHITexture* ColorRT, ERenderTargetActions ColorAction, FRHITexture* DepthRT, EDepthStencilTargetActions DepthActions, FExclusiveDepthStencil InEDS = FExclusiveDepthStencil::DepthWrite_StencilWrite);
    // Color and depth with resolve
    explicit FRHIRenderPassInfo(FRHITexture* ColorRT, ERenderTargetActions ColorAction, FRHITexture* ResolveColorRT,
        FRHITexture* DepthRT, EDepthStencilTargetActions DepthActions, FRHITexture* ResolveDepthRT, FExclusiveDepthStencil InEDS = FExclusiveDepthStencil::DepthWrite_StencilWrite);
    // Color and depth with resolve and optional sample density
    explicit FRHIRenderPassInfo(FRHITexture* ColorRT, ERenderTargetActions ColorAction, FRHITexture* ResolveColorRT,
        FRHITexture* DepthRT, EDepthStencilTargetActions DepthActions, FRHITexture* ResolveDepthRT, FRHITexture* InFoveationTexture, FExclusiveDepthStencil InEDS = FExclusiveDepthStencil::DepthWrite_StencilWrite);

    enum ENoRenderTargets
    {
        NoRenderTargets,
    };
    explicit FRHIRenderPassInfo(ENoRenderTargets Dummy);
    explicit FRHIRenderPassInfo();

    inline int32 GetNumColorRenderTargets() const;
    RHI_API void Validate() const;
    RHI_API void ConvertToRenderTargetsInfo(FRHISetRenderTargetsInfo& OutRTInfo) const;

    (......)
};

// Engine\Source\Runtime\RHI\Public\RHICommandList.h

class RHI_API FRHICommandList : public FRHIComputeCommandList
{
public:
    void BeginRenderPass(const FRHIRenderPassInfo& InInfo, const TCHAR* Name)
    {
        if (InInfo.bTooManyUAVs)
        {
            UE_LOG(LogRHI, Warning, TEXT("RenderPass %s has too many UAVs"));
        }
        InInfo.Validate();

        // 直接调用RHI的接口.
        if (Bypass())
        {
            GetContext().RHIBeginRenderPass(InInfo, Name);
        }
        // 分配RHI命令.
        else
        {
            TCHAR* NameCopy  = AllocString(Name);
            ALLOC_COMMAND(FRHICommandBeginRenderPass)(InInfo, NameCopy);
        }
        // 设置在RenderPass内标记.
        Data.bInsideRenderPass = true;

        // 缓存活动的RT.
        CacheActiveRenderTargets(InInfo);
        // 重置子Pass.
        ResetSubpass(InInfo.SubpassHint);
        Data.bInsideRenderPass = true;
    }

    void EndRenderPass()
    {
        // 调用或分配RHI接口.
        if (Bypass())
        {
            GetContext().RHIEndRenderPass();
        }
        else
        {
            ALLOC_COMMAND(FRHICommandEndRenderPass)();
        }
        // 重置在RenderPass内标记.
        Data.bInsideRenderPass = false;
        // 重置子Pass标记为None.
        ResetSubpass(ESubpassHint::None);
    }
};

它们的使用案例如下：主要是FRHIRenderPassInfo RenderPassInfo(1, ColorRTs, ERenderTargetActions::DontLoad_DontStore)与RHICmdList.BeginRenderPass(RenderPassInfo, TEXT("Test_MultiDrawIndirect"))

bool FRHIDrawTests::Test_MultiDrawIndirect(FRHICommandListImmediate& RHICmdList)
{
	if (!GRHIGlobals.SupportsMultiDrawIndirect)
	{
		return true;
	}

	// Probably could/should automatically enable in the outer scope when running RHI Unit Tests
	// RenderCaptureInterface::FScopedCapture RenderCapture(true /*bEnable*/, &RHICmdList, TEXT("Test_MultiDrawIndirect"));

	static constexpr uint32 MaxInstances = 8;

	// D3D12 does not have a way to get the base instance ID (SV_InstanceID always starts from 0), so we must emulate it...
	const uint32 InstanceIDs[MaxInstances] = { 0, 1, 2, 3, 4, 5, 6, 7 };
	FBufferRHIRef InstanceIDBuffer = CreateBufferWithData(EBufferUsageFlags::VertexBuffer, ERHIAccess::VertexOrIndexBuffer, TEXT("Test_MultiDrawIndirect_InstanceID"), MakeArrayView(InstanceIDs));

	FVertexDeclarationElementList VertexDeclarationElements;
	VertexDeclarationElements.Add(FVertexElement(0, 0, VET_UInt, 0, 4, true /*per instance frequency*/));
	FVertexDeclarationRHIRef VertexDeclarationRHI = PipelineStateCache::GetOrCreateVertexDeclaration(VertexDeclarationElements);

	const uint16 Indices[3] = { 0, 1, 2 };
	FBufferRHIRef IndexBuffer = CreateBufferWithData(EBufferUsageFlags::IndexBuffer, ERHIAccess::VertexOrIndexBuffer, TEXT("Test_MultiDrawIndirect_IndexBuffer"), MakeArrayView(Indices));

	static constexpr uint32 OutputBufferStride = sizeof(uint32);
	static constexpr uint32 OutputBufferSize = OutputBufferStride * MaxInstances;
	FRHIResourceCreateInfo OutputBufferCreateInfo(TEXT("Test_MultiDrawIndirect_OutputBuffer"));
	FBufferRHIRef OutputBuffer = RHICmdList.CreateBuffer(OutputBufferSize, EBufferUsageFlags::UnorderedAccess | EBufferUsageFlags::SourceCopy, OutputBufferStride, ERHIAccess::UAVCompute, OutputBufferCreateInfo);

	const uint32 CountValues[4] = { 1, 1, 16, 0 };
	FBufferRHIRef CountBuffer = CreateBufferWithData(EBufferUsageFlags::DrawIndirect | EBufferUsageFlags::UnorderedAccess, ERHIAccess::IndirectArgs, TEXT("Test_MultiDrawIndirect_Count"), MakeArrayView(CountValues));

	const FRHIDrawIndexedIndirectParameters DrawArgs[] =
	{
		// IndexCountPerInstance, InstanceCount, StartIndexLocation, BaseVertexLocation, StartInstanceLocation
		{3, 1, 0, 0, 0}, // fill slot 0
		// gap in slot 1
		{3, 2, 0, 0, 2}, // fill slots 2, 3 using 1 sub-draw
		// gap in slot 4
		{3, 1, 0, 0, 5}, // fill slots 5, 6 using 2 sub-draws
		{3, 1, 0, 0, 6},
		{3, 1, 0, 0, 7}, // this draw is expected to never execute
	};

	const uint32 ExpectedDrawnInstances[MaxInstances] = { 1, 0, 1, 1, 0, 1, 1, 0 };

	FBufferRHIRef DrawArgBuffer = CreateBufferWithData(EBufferUsageFlags::DrawIndirect | EBufferUsageFlags::UnorderedAccess | EBufferUsageFlags::VertexBuffer, ERHIAccess::IndirectArgs,
		TEXT("Test_MultiDrawIndirect_DrawArgs"), MakeArrayView(DrawArgs));

	FUnorderedAccessViewRHIRef OutputBufferUAV = RHICmdList.CreateUnorderedAccessView(OutputBuffer, 
		FRHIViewDesc::CreateBufferUAV()
		.SetType(FRHIViewDesc::EBufferType::Typed)
		.SetFormat(PF_R32_UINT));

	RHICmdList.ClearUAVUint(OutputBufferUAV, FUintVector4(0));

	const FIntPoint RenderTargetSize(4, 4);
	FRHITextureDesc RenderTargetTextureDesc(ETextureDimension::Texture2D, ETextureCreateFlags::RenderTargetable, PF_B8G8R8A8, FClearValueBinding(), RenderTargetSize, 1, 1, 1, 1, 0);
	FRHITextureCreateDesc RenderTargetCreateDesc(RenderTargetTextureDesc, ERHIAccess::RTV, TEXT("Test_MultiDrawIndirect_RenderTarget"));
	FTextureRHIRef RenderTarget = RHICreateTexture(RenderTargetCreateDesc);

	TShaderMapRef<FTestDrawInstancedVS> VertexShader(GetGlobalShaderMap(GMaxRHIFeatureLevel));
	TShaderMapRef<FTestDrawInstancedPS> PixelShader(GetGlobalShaderMap(GMaxRHIFeatureLevel));

	FGraphicsPipelineStateInitializer GraphicsPSOInit;
	
	GraphicsPSOInit.BoundShaderState.VertexShaderRHI = VertexShader.GetVertexShader();
	GraphicsPSOInit.BoundShaderState.VertexDeclarationRHI = VertexDeclarationRHI;
	GraphicsPSOInit.BoundShaderState.PixelShaderRHI = PixelShader.GetPixelShader();
	GraphicsPSOInit.DepthStencilState = TStaticDepthStencilState<false, CF_Always>::GetRHI();
	GraphicsPSOInit.BlendState = TStaticBlendState<>::GetRHI();
	GraphicsPSOInit.RasterizerState = TStaticRasterizerState<>::GetRHI();
	GraphicsPSOInit.PrimitiveType = EPrimitiveType::PT_TriangleList;

	FRHITexture* ColorRTs[1] = { RenderTarget.GetReference() };
	FRHIRenderPassInfo RenderPassInfo(1, ColorRTs, ERenderTargetActions::DontLoad_DontStore);

	RHICmdList.Transition(FRHITransitionInfo(OutputBufferUAV, ERHIAccess::UAVCompute, ERHIAccess::UAVGraphics, EResourceTransitionFlags::None));
	RHICmdList.BeginUAVOverlap(); // Output UAV can be written without syncs between draws (each draw is expected to write into different slots)

	RHICmdList.BeginRenderPass(RenderPassInfo, TEXT("Test_MultiDrawIndirect"));
	RHICmdList.SetViewport(0, 0, 0, float(RenderTargetSize.X), float(RenderTargetSize.Y), 1);

	RHICmdList.ApplyCachedRenderTargets(GraphicsPSOInit);
	SetGraphicsPipelineState(RHICmdList, GraphicsPSOInit, 0);

	check(InstanceIDBuffer->GetStride() == 4);
	RHICmdList.SetStreamSource(0, InstanceIDBuffer, 0);

	FRHIBatchedShaderParameters ShaderParameters;
	ShaderParameters.SetUAVParameter(PixelShader->OutDrawnInstances.GetBaseIndex(), OutputBufferUAV);
	RHICmdList.SetBatchedShaderParameters(PixelShader.GetPixelShader(), ShaderParameters);

	const uint32 DrawArgsStride = sizeof(DrawArgs[0]);
	const uint32 CountStride = sizeof(CountValues[0]);

	RHICmdList.MultiDrawIndexedPrimitiveIndirect(IndexBuffer, 
		DrawArgBuffer, DrawArgsStride*0, // 1 sub-draw with instance index 0
		CountBuffer, CountStride*0, // count buffer contains 1 in this slot
		5 // expect to draw only 1 instance due to GPU-side upper bound
	);

	RHICmdList.MultiDrawIndexedPrimitiveIndirect(IndexBuffer,
		DrawArgBuffer, DrawArgsStride*1, // 1 sub-draw with 2 instances at base index 2
		CountBuffer, CountStride*1, // count buffer contains 1 in this slot
		4 // expect to draw only 1 instance due to GPU-side upper bound
	);

	RHICmdList.MultiDrawIndexedPrimitiveIndirect(IndexBuffer,
		DrawArgBuffer, DrawArgsStride*2, // 2 sub-draws with 1 instance each starting at base index 5
		CountBuffer, CountStride*2, // count buffer contains 16 in this slot
		2 // expect to draw only 2 instances due to CPU-side upper bound
	);

	RHICmdList.MultiDrawIndexedPrimitiveIndirect(IndexBuffer,
		DrawArgBuffer, DrawArgsStride*4, // 1 sub-draw with 1 instance each starting at base index 7
		CountBuffer, CountStride*3, // count buffer contains 0 in this slot
		1 // expect to skip the draw due to GPU-side count of 0
	);

	RHICmdList.MultiDrawIndexedPrimitiveIndirect(IndexBuffer,
		DrawArgBuffer, DrawArgsStride*4, // 1 sub-draw with 1 instance each starting at base index 7
		CountBuffer, CountStride*0, // count buffer contains 1 in this slot
		0 // expect to skip the draw due to CPU-side count of 0
	);

	RHICmdList.EndRenderPass();

	RHICmdList.EndUAVOverlap();

	RHICmdList.Transition(FRHITransitionInfo(OutputBufferUAV, ERHIAccess::UAVGraphics, ERHIAccess::CopySrc, EResourceTransitionFlags::None));

	TConstArrayView<uint8> ExpectedOutputView = MakeArrayView(reinterpret_cast<const uint8*>(ExpectedDrawnInstances), sizeof(ExpectedDrawnInstances));
	bool bSucceeded = FRHIBufferTests::VerifyBufferContents(TEXT("Test_MultiDrawIndirect"), RHICmdList, OutputBuffer, ExpectedOutputView);

	return bSucceeded;
}

Subpass

先说一下Subpass的由来、作用和特点。

在传统的多Pass渲染中，每个Pass结束时通常会渲染出一组渲染纹理，部分成为着色器参数提供给下一个Pass采样读取。这种纹理采样方式不受任何限制，可以读取任意的领域像素，使用任意的纹理过滤方式。这种方式虽然使用灵活，但在TBR（Tile-Based Renderer）硬件架构的设备中会有较大的消耗：渲染纹理的Pass通常会将渲染结果存储在On-chip的Tile Memory中，待Pass结束后会写回GPU显存（VRAM）中，写回GPU显存是个耗时耗耗电的操作。

传统多Pass之间的内存存取模型，多次发生于On-Chip和全局存储器之间。

如果出现一种特殊的纹理使用情况：上一个Pass渲染处理的纹理，立即被下一个Pass使用，并且下一个Pass只采样像素位置自身的数据，而不需要采样邻域像素的位置。这种情况就符合了Subpass的使用情景。使用Subpass渲染的纹理结果只会存储在Tile Memory中，在Subpass结束后不会写回VRAM，而直接提供Tile Memory的数据给下一个Subpass采样读取。这样就避免了传统Pass结束写回GPU显存以及下一个Pass又从GPU显存读数据的耗时耗电操作，从而提升了性能。

Subpass之间的内存存取模型，都发生在On-Chip内。

Subpass的相关代码主要集中在移动端中。UE涉及Subpass的接口和类型如下：

// 提供给RHI的Subpass标记.
enum class ESubpassHint : uint8
{
    None,                    // 传统渲染(非Subpass)
    DepthReadSubpass,        // 深度读取Subpass.
    DeferredShadingSubpass, // 移动端延迟着色Subpass.
};


// Engine\Source\Runtime\RHI\Public\RHICommandList.h

class RHI_API FRHICommandListBase : public FNoncopyable
{
    (......)
    
protected:
    // PSO上下文.
    struct FPSOContext
    {
        uint32 CachedNumSimultanousRenderTargets = 0;
        TStaticArray<FRHIRenderTargetView, MaxSimultaneousRenderTargets> CachedRenderTargets;
        FRHIDepthRenderTargetView CachedDepthStencilTarget;
        
        // Subpass提示标记.
        ESubpassHint SubpassHint = ESubpassHint::None;
        uint8 SubpassIndex = 0;
        uint8 MultiViewCount = 0;
        bool HasFragmentDensityAttachment = false;
    } PSOContext;
};

class RHI_API FRHICommandList : public FRHIComputeCommandList
{
public:
    void BeginRenderPass(const FRHIRenderPassInfo& InInfo, const TCHAR* Name)
    {
        (......)

        CacheActiveRenderTargets(InInfo);
        // 设置Subpass数据.
        ResetSubpass(InInfo.SubpassHint);
        Data.bInsideRenderPass = true;
    }

    void EndRenderPass()
    {
        (......)
        
        // 重置Subpass标记为None.
        ResetSubpass(ESubpassHint::None);
    }

    // 下一个Subpass.
    void NextSubpass()
    {
        // 分配或调用RHI接口.
        if (Bypass())
        {
            GetContext().RHINextSubpass();
        }
        else
        {
            ALLOC_COMMAND(FRHICommandNextSubpass)();
        }
        
        // 增加Subpass计数.
        IncrementSubpass();
    }
    
    // 增加subpass计数.
    void IncrementSubpass()
    {
        PSOContext.SubpassIndex++;
    }
    
    // 重置Subpass数据.
    void ResetSubpass(ESubpassHint SubpassHint)
    {
        PSOContext.SubpassHint = SubpassHint;
        PSOContext.SubpassIndex = 0;
    }
};

36 KiB Raw Blame History Unescape Escape

前言

概念

FDynamicRHI

FParallelCommandListSet

普通Pass渲染

Subpass

36 KiB

Raw Blame History