https://github.com/ConfettiFX/The-Forge/blob/master/CommonRaytracing_3/ThirdParty/DXR/doc/D3D12%20Raytracing%20Functional%20Spec.docx
dxr的spec 好难找
D3D12 Raytracing Functional Spec
https://github.com/NVIDIAGameWorks/Falcor/releases
https://github.com/Microsoft/DirectX-Graphics-Samples/blob/master/Samples/Desktop/D3D12Raytracing/readme.md
https://github.com/Microsoft/DirectX-Graphics-Samples/blob/master/Libraries/D3D12RaytracingFallback/readme.md
http://forums.directxtech.com/index.php?topic=5860.0
初始化拿到device 对rtx开一些特殊设置
fallback是个模拟rtx的库 api和硬件的rtx不太一样要专门写
初始化各种资源 初始化rtx专有资源
pso
as
加速结构 top bottom
top是对bottom的一个transform 数据本身在bottom
可以有一组bottome 每个里面又可以有很多geometry
语法参见 bottomLevelBuildDesc.NumDescs = static_cast<UINT>(geometryDescs.size());
它公用dx12的cmd自己再这个cmd上面建一个rtx的cmd
有五种shader可以写
hit 里面的光照是 光线打到的物体上的光照 比如反射 了abc 是abc物体的光照 而不是反射平面本身的材质光照
从传统光栅化shader到这种hit的光照shader的转换
vs部分的pos normal uv之类的 从hit的结果能拿到
只port ps光照部分即可 返回光照结果
几天没见pix长进了
https://blogs.msdn.microsoft.com/pix/gpu-captures/
看上去不错
https://blogs.msdn.microsoft.com/pix/2018/07/24/pix-1807-19-shader-table-viewer/
dx12和dxr都可以debug 需要把dxr的dll copy到pix.exe的位置
gpu hang的debug
https://blogs.msdn.microsoft.com/pix/tdr-debugging/
https://docs.microsoft.com/zh-cn/windows/desktop/direct3d12/using-d3d12-debug-layer-gpu-based-validation
https://blogs.msdn.microsoft.com/pix/programmatic-capture/
很强大
https://blogs.msdn.microsoft.com/pix/remoting/
远程连接
---------------------------
dx12 rootsignature ---SRV CBV UAV 初始到各个slot
m_device->CreateConstantBufferView(&cbvDesc, m_cbvHeap->GetCPUDescriptorHandleForHeapStart()); 用这句连
CB--View --descriptorHeap
init 的时候把这些连上
render的时候 用这种方式更新 m_commandList->SetGraphicsRootDescriptorTable(0, m_cbvHeap->GetGPUDescriptorHandleForHeapStart()); 用heap更新slot
update更新了cb本身
dxr里面有shadertable 是upload buffer--committed resource
一条layout是一个shader rootsig mat等等
auto DispatchRays = [&](auto* commandList, auto* stateObject, auto* dispatchDesc)
{
dispatchDesc->HitGroupTable.StartAddress = m_hitGroupShaderTable->GetGPUVirtualAddress();
dispatchDesc->HitGroupTable.SizeInBytes = m_hitGroupShaderTable->GetDesc().Width;
dispatchDesc->HitGroupTable.StrideInBytes = m_hitGroupShaderTableStrideInBytes;
dispatchDesc->MissShaderTable.StartAddress = m_missShaderTable->GetGPUVirtualAddress();
dispatchDesc->MissShaderTable.SizeInBytes = m_missShaderTable->GetDesc().Width;
dispatchDesc->MissShaderTable.StrideInBytes = m_missShaderTableStrideInBytes;
dispatchDesc->RayGenerationShaderRecord.StartAddress = m_rayGenShaderTable->GetGPUVirtualAddress();
dispatchDesc->RayGenerationShaderRecord.SizeInBytes = m_rayGenShaderTable->GetDesc().Width;
dispatchDesc->Width = m_width;
dispatchDesc->Height = m_height;
commandList->DispatchRays(stateObject, dispatchDesc);
};
DispatchRays 可以把三种shader的起始地址连上
cbv之类的更新同dx12
namespace RootSignatureSlots = LocalRootSignature::AABB::Slot;
CD3DX12_ROOT_PARAMETER rootParameters[RootSignatureSlots::Count];
rootParameters[RootSignatureSlots::MaterialConstant].InitAsConstants(SizeOfInUint32(PrimitiveConstantBuffer), 1);
rootParameters[RootSignatureSlots::GeometryIndex].InitAsConstants(SizeOfInUint32(PrimitiveInstanceConstantBuffer), 2);
ThrowIfFailed(D3D12SerializeRootSignature(&desc, D3D_ROOT_SIGNATURE_VERSION_1, &blob, &error), error ? static_cast<wchar_t*>(error->GetBufferPointer()) : nullptr);
ThrowIfFailed(device->CreateRootSignature(1, blob->GetBufferPointer(), blob->GetBufferSize(), IID_PPV_ARGS(&(*rootSig))));
建立的时候 每条shaderrecord里面matID是不一样的 在hitshader里面
uint materialID = MaterialID;
uint triangleID = PrimitiveIndex();
RayTraceMeshInfo info = g_meshInfo[materialID];
const uint3 ii = Load3x16BitIndices(info.m_indexOffsetBytes + PrimitiveIndex() * 3 * 2);
const float2 uv0 = GetUVAttribute(info.m_uvAttributeOffsetBytes + ii.x * info.m_attributeStrideBytes);
const float2 uv1 = GetUVAttribute(info.m_uvAttributeOffsetBytes + ii.y * info.m_attributeStrideBytes);
const float2 uv2 = GetUVAttribute(info.m_uvAttributeOffsetBytes + ii.z * info.m_attributeStrideBytes);
就可以用这种方式matID拿到具体数据
==================
hit之后的顶点信息有两种方式可以获取 gloable /local sig
gloable的方式 需要传matID每次用这个ID从大的buffer(vb包含所有mesh)作为偏移拿到具体 pos uv normal tangent信息
另一种local的方式 每个hit shader传 自己的vb数据(initialze的时候每块mesh来处理 生成不同的 shader和vb)作为cb传入 用hit时的primitiveIndex拿到相应数据
这样hit的mat也都是每块mesh不一样的了
==========
这是些intric val
Values \ shaders |
Ray dispatch system values: |
uint2 DispatchRaysIndex() |
uint2 DispatchRaysDimensions() |
Ray system values: |
float3 WorldRayOrigin() |
float3 WorldRayDirection() |
float RayTMin() |
float RayTCurrent() |
uint RayFlags() |
Primitive/object space system values: |
uint InstanceIndex() |
uint InstanceID() |
uint PrimitiveIndex() |
float3 ObjectRayOrigin() |
float3 ObjectRayDirection() |
float3x4 ObjectToWorld() |
float3x4 WorldToObject() |
Hit specific system values: |
uint HitKind() |
=========================
draw() rastization
dispatch() compute
dispatchRays() raytracing
invoke