Use buffer device address.

Assets/Shaders/Mesh.slang

@@ -2,8 +2,8 @@ import Bindless;
 
 struct VertexOut {
     float4 outPosition      : SV_Position;
-    float4 screenPosition   : ScreenPosition;
+    float4 worldPosition    : WorldPosition;
-    float4 normal           : CoarseNormal;
+    float4 normal           : WorldNormal;
     float2 texCoord0        : TexCoord0;
     float2 texCoord1        : TexCoord1;
     float4 vertexColor0     : VertexColor;
@@ -12,9 +12,46 @@ struct VertexOut {
 struct CameraData {
     float4x4 view;
     float4x4 proj;
+    float4 position;
 };
 
-uniform ParameterBlock<CameraData> camera;
+struct PointLight {
+    float3 position;
+    float  range;
+    float3 color;
+    float  attenuation;
+};
+
+struct DirectionalLight {
+    float3 direction;
+    float  _padding0;
+    float3 color;
+    float  _padding1;
+};
+
+struct LightData {
+    PointLight* pointLights;
+    DirectionalLight* dirLights;
+    uint pointLightCount;
+    uint dirLightCount;
+
+    PointLight getPointLight(uint idx) {
+        if (idx >= pointLightCount) return pointLights[0]; 
+        return pointLights[idx];
+    }
+
+    DirectionalLight getDirectionalLight(uint idx) {
+        if (idx >= dirLightCount) return dirLights[0]; 
+        return dirLights[idx];
+    }
+};
+
+struct PerFrameData {
+    CameraData camera;
+    LightData lightData;
+};
+
+uniform ParameterBlock<PerFrameData> pfd;
 
 struct PerInstanceData {
     float4x4 transform;
@@ -37,9 +74,11 @@ VertexOut VertexMain(
     float2 texCoord1,
     float4 vertexColor0,
 ) {
+    float4 worldPosition = mul(pcb.transform, float4(position, 1.0f));
+
     VertexOut output;
-    output.outPosition = mul(camera.proj, mul(camera.view, mul(pcb.transform, float4(position, 1.0f))));
+    output.outPosition = mul(pfd.camera.proj, mul(pfd.camera.view, worldPosition));
-    output.screenPosition = mul(camera.proj, mul(camera.view, mul(pcb.transform, float4(position, 1.0f))));
+    output.worldPosition = worldPosition;
     output.normal = mul(pcb.transform, float4(normalize(normal.rgb), 0.0f));
     output.texCoord0 = texCoord0;
     output.texCoord1 = texCoord1;
@@ -49,20 +88,40 @@ VertexOut VertexMain(
 
 [shader("fragment")]
 float4 FragmentMain(
-    float4 interpolatePosition  : ScreenPosition,
+    float4 worldPosition : WorldPosition,
-    float4 interpolatedNormal   : CoarseNormal,
+    float4 normal       : WorldNormal,
-    float2 uv0                  : TexCoord0,
+    float2 uv0          : TexCoord0,
-    float2 uv1                  : TexCoord1,
+    float2 uv1          : TexCoord1,
-    float4 interpolatedColor    : VertexColor,
+    float4 color        : VertexColor,
 ) : SV_Target0 {
 
-    let N = interpolatedNormal.xyz;
+    float3 diffuse = 0.0f.xxx;
-    let L = dot(normalize(N), float3(1.0f, 1.0f, -1.0f));
+    float3 specular = 0.0f.xxx;
+
+    for (uint i = 0; i < pfd.lightData.pointLightCount; ++i) {
+        PointLight pointlight = pfd.lightData.pointLights[i];
+
+        let lightPosition = pointlight.position;
+        let lightDisplace = worldPosition.xyz - lightPosition;
+        let lightDistance = length(lightDisplace);
+        let lightDirection = normalize(lightDisplace);
+        let viewDirection = normalize(worldPosition.xyz - pfd.camera.position.xyz);
+        let halfWayVector = normalize(-lightDirection + viewDirection);
+
+        let attenuation = (1.0f / lightDistance);
+
+        let diffuseFactor = pcb.roughness * dot(-lightDirection, normalize(normal.xyz));
+        diffuse += pointlight.color * diffuseFactor;
+
+        let specularFactor = (1.0f - pcb.roughness) * pow(max(dot(halfWayVector, viewDirection), 0.0f), 32.0f) * attenuation;
+
+        specular += pointlight.color * specularFactor;
+    }
 
     if (let texture = pcb.textureID) {
-        return float4(texture.Sample(uv0).rgb, 1.0f) * pcb.baseColor * interpolatedColor * L;
+        return float4(texture.Sample(uv0).rgb, 1.0f) * pcb.baseColor * color * float4((diffuse + specular), 0.0f);
     } else {
-        return pcb.baseColor * interpolatedColor * L;
+        return pcb.baseColor * color * float4((diffuse + specular), 0.0f);
     }
 }
 

Blaze.vcxproj

@@ -119,7 +119,7 @@
     <Link>
       <SubSystem>Console</SubSystem>
       <GenerateDebugInformation>true</GenerateDebugInformation>
-      <AdditionalLibraryDirectories>C:\Users\Eon\source\repos\Blaze\vcpkg_installed\x64-windows\x64-windows\bin;%(AdditionalLibraryDirectories)</AdditionalLibraryDirectories>
+      <AdditionalLibraryDirectories>%(AdditionalLibraryDirectories)</AdditionalLibraryDirectories>
     </Link>
   </ItemDefinitionGroup>
   <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
@@ -144,7 +144,7 @@
       <EnableCOMDATFolding>true</EnableCOMDATFolding>
       <OptimizeReferences>true</OptimizeReferences>
       <GenerateDebugInformation>true</GenerateDebugInformation>
-      <AdditionalLibraryDirectories>C:\Users\Eon\source\repos\Blaze\vcpkg_installed\x64-windows\x64-windows\bin;%(AdditionalLibraryDirectories)</AdditionalLibraryDirectories>
+      <AdditionalLibraryDirectories>%(AdditionalLibraryDirectories)</AdditionalLibraryDirectories>
     </Link>
   </ItemDefinitionGroup>
   <ItemGroup>

Blaze/Blaze.cpp

@@ -50,6 +50,37 @@ SDL_AppResult SDL_AppInit( void** appstate, int, char** )
       LoadModel( appState.renderDevice, appState.entityManager, "Assets/Models/OrientationTest.glb" );
   ASSERT( entity );
 
+  std::array pointLight = {
+    MiscData::PointLight{
+                         .position    = { 12.0f, 0.0f, 0.0f },
+                         .range       = 12,
+                         .color       = { 1.0f, 0.0f, 0.0f },
+                         .attenuation = 1.0f,
+                         },
+    MiscData::PointLight{
+                         .position    = { 0.0f, 12.0f, 0.0f },
+                         .range       = 12,
+                         .color       = { 0.0f, 1.0f, 0.0f },
+                         .attenuation = 1.0f,
+                         },
+    MiscData::PointLight{
+                         .position    = { 0.0f, 0.0f, -12.0f },
+                         .range       = 6,
+                         .color       = { 0.0f, 0.0f, 1.0f },
+                         .attenuation = 1.0f,
+                         },
+  };
+
+  appState.miscData->lightData.pointLightCount = static_cast<uint32_t>( pointLight.size() );
+
+  appState.renderDevice->bufferManager->writeToBuffer(
+      appState.miscData->pointLights, std::span{ pointLight.begin(), pointLight.end() } );
+
+  memcpy(
+      appState.miscData->cameraUniformBufferPtr + sizeof( MiscData::CameraData ),
+      &appState.miscData->lightData,
+      sizeof appState.miscData->lightData );
+
   return SDL_APP_CONTINUE;
 }
 
@@ -260,7 +291,6 @@ SDL_AppResult SDL_AppIterate( void* appstate )
                 materialData );
 
             vkCmdDrawIndexed( cmd, primitive.indexCount, 1, primitive.indexStart, primitive.vertexOffset, 0 );
-            // vkCmdDrawIndexed( cmd, primitive.count, 1, primitive.start, 0 );
           }
         }
 

Blaze/BufferManager.cpp

@@ -46,7 +46,7 @@ void BufferManager::writeToBufferImpl( BufferID const& rid, void const* data, si
 
   Buffer const& buffer = fetchBufferUnchecked( rid );
 
-  ASSERT( buffer.size <= size );
+  ASSERT( size <= buffer.size );
 
   memcpy( buffer.mappedData, data, size );
 }
@@ -112,11 +112,12 @@ std::optional<BufferID> BufferManager::createVertexBuffer( size_t const size )
   // NOTE: textureSlot preserves index between uses.
   uint32_t index = bufferSlot->index;
   new ( bufferSlot ) Buffer{
-    .buffer     = vertexBuffer,
+    .buffer        = vertexBuffer,
-    .allocation = vertexBufferAllocation,
+    .allocation    = vertexBufferAllocation,
-    .mappedData = static_cast<std::byte*>( allocationInfo.pMappedData ),
+    .mappedData    = static_cast<std::byte*>( allocationInfo.pMappedData ),
-    .size       = size,
+    .deviceAddress = 0,
-    .index      = index,
+    .size          = size,
+    .index         = index,
   };
 
   // NOTE: Memory hackery to create TextureID;
@@ -174,11 +175,83 @@ std::optional<BufferID> BufferManager::createIndexBuffer( size_t size )
   // NOTE: bufferSlot preserves index between uses.
   uint32_t index = bufferSlot->index;
   new ( bufferSlot ) Buffer{
-    .buffer     = indexBuffer,
+    .buffer        = indexBuffer,
-    .allocation = indexBufferAllocation,
+    .allocation    = indexBufferAllocation,
-    .mappedData = static_cast<std::byte*>( allocationInfo.pMappedData ),
+    .mappedData    = static_cast<std::byte*>( allocationInfo.pMappedData ),
-    .size       = size,
+    .deviceAddress = 0,
-    .index      = index,
+    .size          = size,
+    .index         = index,
+  };
+
+  // NOTE: Memory hackery to create BufferID;
+  return std::move( *reinterpret_cast<BufferID*>( &index ) );
+}
+
+std::optional<BufferID> BufferManager::createStorageBuffer( size_t size )
+{
+  if ( m_freeList.empty() )
+  {
+    return std::nullopt;
+  }
+
+  Buffer* bufferSlot = reinterpret_cast<Buffer*>( m_freeList.popFront() );
+  ++m_count;
+
+  ASSERT( m_pRenderDevice );
+  RenderDevice const&      renderDevice     = *m_pRenderDevice;
+
+
+  VkBufferCreateInfo const bufferCreateInfo = {
+    .sType                 = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
+    .pNext                 = nullptr,
+    .flags                 = 0,
+    .size                  = size,
+    .usage                 = VK_BUFFER_USAGE_STORAGE_BUFFER_BIT | VK_BUFFER_USAGE_SHADER_DEVICE_ADDRESS_BIT,
+    .sharingMode           = VK_SHARING_MODE_EXCLUSIVE,
+    .queueFamilyIndexCount = 0,
+    .pQueueFamilyIndices   = nullptr,
+  };
+
+  VmaAllocationCreateInfo constexpr allocationCreateInfo = {
+    .flags          = VMA_ALLOCATION_CREATE_MAPPED_BIT | VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT,
+    .usage          = VMA_MEMORY_USAGE_AUTO,
+    .requiredFlags  = VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT,
+    .preferredFlags = 0,
+    .memoryTypeBits = 0,
+    .pool           = nullptr,
+    .pUserData      = nullptr,
+    .priority       = 1.0f,
+  };
+
+  VmaAllocationInfo allocationInfo;
+  VkBuffer          storageBuffer;
+  VmaAllocation     storageBufferAllocation;
+
+  VK_CHECK( vmaCreateBuffer(
+      renderDevice.gpuAllocator,
+      &bufferCreateInfo,
+      &allocationCreateInfo,
+      &storageBuffer,
+      &storageBufferAllocation,
+      &allocationInfo ) );
+
+  VkBufferDeviceAddressInfo const deviceAddressInfo = {
+    .sType  = VK_STRUCTURE_TYPE_BUFFER_DEVICE_ADDRESS_INFO,
+    .pNext  = nullptr,
+    .buffer = storageBuffer,
+  };
+
+  VkDeviceAddress const deviceAddress = vkGetBufferDeviceAddress( renderDevice.device, &deviceAddressInfo );
+
+  // NOTE: bufferSlot preserves index between uses.
+  uint32_t index = bufferSlot->index;
+  new ( bufferSlot ) Buffer{
+    .buffer        = storageBuffer,
+    .allocation    = storageBufferAllocation,
+    .mappedData    = static_cast<std::byte*>( allocationInfo.pMappedData ),
+    .deviceAddress = deviceAddress,
+    .size          = size,
+    .index         = index,
   };
 
   // NOTE: Memory hackery to create BufferID;
@@ -203,6 +276,16 @@ std::optional<VkBuffer> BufferManager::fetchBuffer( BufferID const& rid )
   return fetchBufferUnchecked( rid ).buffer;
 }
 
+std::optional<VkDeviceAddress> BufferManager::fetchDeviceAddress( BufferID const& rid )
+{
+  if ( not isValidID( rid ) ) return std::nullopt;
+  Buffer const& buffer = fetchBufferUnchecked( rid );
+
+  if ( buffer.deviceAddress == 0 ) return std::nullopt;
+
+  return buffer.deviceAddress;
+}
+
 BufferManager::BufferManager( RenderDevice* pRenderDevice, Buffer* aBuffers, uint32_t const capacity )
   : m_pRenderDevice{ pRenderDevice }, m_aBuffers{ aBuffers }, m_count{ 0 }, m_capacity{ capacity }
 {

Blaze/BufferManager.h

@@ -14,11 +14,12 @@ struct RenderDevice;
 
 struct Buffer
 {
-  VkBuffer      buffer;
+  VkBuffer        buffer;
-  VmaAllocation allocation;
+  VmaAllocation   allocation;
-  std::byte*    mappedData; // Assume the system has ReBAR/SAM enabled.
+  std::byte*      mappedData; // Assume the system has ReBAR/SAM enabled.
-  size_t        size;
+  VkDeviceAddress deviceAddress;
-  uint32_t      index;
+  size_t          size;
+  uint32_t        index;
 };
 
 static_assert( sizeof( Buffer ) > sizeof( FreeList::Node ) and "Buffer is used intrusively by FreeList" );
@@ -55,18 +56,21 @@ public:
   [[nodiscard]] bool      isValidID( BufferID const& rid ) const;
 
   std::optional<BufferID> createVertexBuffer( size_t size );
-
   std::optional<BufferID> createIndexBuffer( size_t size );
+  std::optional<BufferID> createStorageBuffer( size_t size );
 
   void                    freeBuffer( BufferID&& rid );
 
   DEPRECATE_JULY_2025
-  std::optional<VkBuffer> fetchBuffer( BufferID const& rid );
+  std::optional<VkBuffer>        fetchBuffer( BufferID const& rid );
+  std::optional<VkDeviceAddress> fetchDeviceAddress( BufferID const& rid );
 
-  template <typename T>
+  void                           writeToBuffer( BufferID const& rid, std::ranges::contiguous_range auto const& data )
-  void writeToBuffer( BufferID const& rid, std::span<T> const& data )
   {
-    writeToBufferImpl( rid, data.data(), data.size_bytes() );
+    writeToBufferImpl(
+        rid,
+        std::ranges::data( data ),
+        std::ranges::size( data ) * sizeof( std::ranges::range_value_t<decltype( data )> ) );
   }
 
   //

Blaze/MiscData.cpp

@@ -43,7 +43,7 @@ bool MiscData::init( RenderDevice const& renderDevice )
       .binding            = 0,
       .descriptorType     = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER,
       .descriptorCount    = 1,
-      .stageFlags         = VK_SHADER_STAGE_VERTEX_BIT,
+      .stageFlags         = VK_SHADER_STAGE_VERTEX_BIT | VK_SHADER_STAGE_FRAGMENT_BIT,
       .pImmutableSamplers = nullptr,
     };
 
@@ -60,7 +60,7 @@ bool MiscData::init( RenderDevice const& renderDevice )
     VkPushConstantRange const pushConstantRange = {
       .stageFlags = VK_SHADER_STAGE_ALL_GRAPHICS,
       .offset     = 0,
-      .size       = sizeof( DirectX::XMMATRIX ) + Material::GPU_DATA_SIZE
+      .size       = sizeof( DirectX::XMMATRIX ) + Material::GPU_DATA_SIZE,
     };
 
     std::array const descriptorSetLayouts = {
@@ -289,14 +289,37 @@ bool MiscData::init( RenderDevice const& renderDevice )
 
   // Camera
   {
-    cameraPosition        = DirectX::XMVectorSet( 0.0f, 20.0f, -20.0f, 1.0f );
+    cameraData.cameraPosition = DirectX::XMVectorSet( 0.0f, 20.0f, -20.0f, 1.0f );
-    cameraTarget          = DirectX::XMVectorSet( 0.0f, 0.0f, 0.0f, 1.0f );
+    cameraTarget              = DirectX::XMVectorSet( 0.0f, 0.0f, 0.0f, 1.0f );
-    cameraUp              = DirectX::XMVectorSet( 0.0f, 1.0f, 0.0f, 1.0f );
+    cameraUp                  = DirectX::XMVectorSet( 0.0f, 1.0f, 0.0f, 1.0f );
-    cameraData.viewMatrix = DirectX::XMMatrixLookAtLH( cameraPosition, cameraTarget, cameraUp );
+    cameraData.viewMatrix     = DirectX::XMMatrixLookAtLH( cameraData.cameraPosition, cameraTarget, cameraUp );
     cameraData.projectionMatrix =
         DirectX::XMMatrixPerspectiveFovLH( DirectX::XMConvertToRadians( 70.0f ), 16.0f / 9.0f, 0.1f, 1000.0f );
 
-    cameraUniformBufferSize                   = sizeof( CameraData );
+    cameraUniformBufferSize = sizeof( CameraData ) + sizeof( LightData );
+  }
+
+
+  // Lights
+  {
+    auto pointLightsValue = renderDevice.bufferManager->createStorageBuffer( 10 * sizeof( PointLight ) );
+    if ( !pointLightsValue ) return false;
+
+    pointLights         = std::move( pointLightsValue.value() );
+
+    auto dirLightsValue = renderDevice.bufferManager->createStorageBuffer( 10 * sizeof( DirectionalLight ) );
+    if ( !dirLightsValue ) return false;
+
+    directionalLights           = std::move( dirLightsValue.value() );
+
+    lightData.pointLights       = renderDevice.bufferManager->fetchDeviceAddress( pointLights ).value();
+    lightData.directionalLights = renderDevice.bufferManager->fetchDeviceAddress( directionalLights ).value();
+    lightData.dirLightCount     = 0;
+    lightData.pointLightCount   = 0;
+  }
+
+  // Uniform Buffer
+  {
 
     VkBufferCreateInfo const bufferCreateInfo = {
       .sType                 = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
@@ -332,8 +355,9 @@ bool MiscData::init( RenderDevice const& renderDevice )
 
     if ( allocationInfo.pMappedData )
     {
-      memcpy( allocationInfo.pMappedData, &cameraData, sizeof cameraData );
       cameraUniformBufferPtr = static_cast<uint8_t*>( allocationInfo.pMappedData );
+      memcpy( cameraUniformBufferPtr, &cameraData, sizeof cameraData );
+      memcpy( cameraUniformBufferPtr + sizeof cameraData, &lightData, sizeof lightData );
     }
   }
 
@@ -373,7 +397,7 @@ bool MiscData::init( RenderDevice const& renderDevice )
     VkDescriptorBufferInfo const descriptorBufferInfo = {
       .buffer = cameraUniformBuffer,
       .offset = 0,
-      .range  = sizeof CameraData,
+      .range  = cameraUniformBufferSize,
     };
 
     std::array writeDescriptorSets = {
@@ -474,6 +498,9 @@ void MiscData::destroy( RenderDevice const& renderDevice )
   vkDestroyDescriptorPool( device, Take( descriptorPool ), nullptr );
   vmaDestroyBuffer( renderDevice.gpuAllocator, Take( cameraUniformBuffer ), Take( cameraUniformBufferAllocation ) );
 
+  renderDevice.bufferManager->freeBuffer( std::move( pointLights ) );
+  renderDevice.bufferManager->freeBuffer( std::move( directionalLights ) );
+
   vkDestroyPipeline( device, Take( meshPipeline ), nullptr );
   vkDestroyPipelineLayout( device, Take( pipelineLayout ), nullptr );
   vkDestroyDescriptorSetLayout( device, Take( descriptorSetLayout ), nullptr );

Blaze/MiscData.h

@@ -6,6 +6,8 @@
 
 #include <DirectXMath.h>
 
+#include "BufferManager.h"
+
 struct GlobalMemory;
 struct RenderDevice;
 
@@ -15,6 +17,31 @@ struct MiscData
   {
     DirectX::XMMATRIX viewMatrix;
     DirectX::XMMATRIX projectionMatrix;
+    DirectX::XMVECTOR cameraPosition;
+  };
+
+  struct PointLight
+  {
+    DirectX::XMFLOAT3 position;
+    float             range;
+    DirectX::XMFLOAT3 color;
+    float             attenuation;
+  };
+
+  struct DirectionalLight
+  {
+    DirectX::XMFLOAT3 direction;
+    float             _padding0;
+    DirectX::XMFLOAT3 color;
+    float             _padding1;
+  };
+
+  struct LightData
+  {
+    VkDeviceAddress pointLights;
+    VkDeviceAddress directionalLights;
+    uint32_t        pointLightCount;
+    uint32_t        dirLightCount;
   };
 
   uint64_t              previousCounter;
@@ -23,10 +50,14 @@ struct MiscData
   VkPipelineLayout      pipelineLayout;
   VkPipeline            meshPipeline;
 
-  DirectX::XMVECTOR     cameraPosition;
   DirectX::XMVECTOR     cameraTarget;
   DirectX::XMVECTOR     cameraUp;
   CameraData            cameraData;
+
+  BufferID              pointLights;
+  BufferID              directionalLights;
+  LightData             lightData;
+
   VkBuffer              cameraUniformBuffer;
   VmaAllocation         cameraUniformBufferAllocation;
   size_t                cameraUniformBufferSize;

Blaze/ModelLoader.cpp

@@ -31,7 +31,6 @@ uint32_t ProcessMaterial( RenderDevice* renderDevice, Model* model, cgltf_materi
     cgltf_image* baseColorImage = material.pbr_metallic_roughness.base_color_texture.texture->image;
 
     {
-
       byte* data;
       if ( baseColorImage->buffer_view->data )
       {
@@ -740,7 +739,7 @@ Entity* LoadModel( RenderDevice* renderDevice, EntityManager* entityManager, con
   }
   entity->model.vertexBuffer = std::move( vertexBuffer.value() );
 
-  renderDevice->bufferManager->writeToBuffer( entity->model.vertexBuffer, std::span{ vertices } );
+  renderDevice->bufferManager->writeToBuffer( entity->model.vertexBuffer, vertices );
 
   auto indexBuffer = renderDevice->bufferManager->createIndexBuffer( indices.size() * sizeof indices[0] );
   if ( not indexBuffer )

Blaze/RenderDevice.cpp

@@ -192,6 +192,7 @@ RenderDevice* RenderDevice_Create( GlobalMemory* mem, RenderDevice::CreateInfo c
       .descriptorBindingPartiallyBound               = true,
       .descriptorBindingVariableDescriptorCount      = true,
       .runtimeDescriptorArray                        = true,
+      .bufferDeviceAddress                           = true,
     };
 
     VkPhysicalDeviceFeatures features = {
@@ -218,7 +219,7 @@ RenderDevice* RenderDevice_Create( GlobalMemory* mem, RenderDevice::CreateInfo c
     volkLoadDevice( device );
 
     VmaAllocatorCreateInfo allocatorCreateInfo = {
-      .flags                          = 0,
+      .flags                          = VMA_ALLOCATOR_CREATE_BUFFER_DEVICE_ADDRESS_BIT,
       .physicalDevice                 = physicalDeviceInUse,
       .device                         = device,
       .preferredLargeHeapBlockSize    = 0,