it works now???

res/shaders/LegacyBasePass.slang

@@ -9,6 +9,6 @@ FragmentParameter vertexMain(
 	uint instanceId: SV_InstanceID,
 ){
 	InstanceData inst = pScene.instances[instanceId];
-	VertexAttributes attr = pVertexData.getAttributes(vertexId, inst.transformMatrix);
+	VertexAttributes attr = pVertexData.getAttributes(vertexId);
 	return attr.getParameter(inst.transformMatrix);
 }

res/shaders/MeshletBasePass.slang

@@ -64,21 +64,11 @@ void taskMain(
 	DispatchMesh(head, 1, 1, p);
 }
 
-groupshared VertexAttributes gs_vertices[MAX_VERTICES];
-groupshared uint3 gs_indices[MAX_PRIMITIVES];
-groupshared uint gs_numVertices;
-groupshared uint gs_numPrimitives;
-
 struct PrimitiveAttributes
 {
 	uint cull: SV_CullPrimitive;
 };
 
-struct MeshOutput
-{
-	FragmentParameter parameter;
-};
-
 [numthreads(MESH_GROUP_SIZE, 1, 1)]
 [outputtopology("triangle")]
 [shader("mesh")]
@@ -86,73 +76,50 @@ void meshMain(
 	in uint threadID: SV_GroupIndex,
 	in uint groupID: SV_GroupID,
 	in payload MeshPayload meshPayload,
-	out Vertices<MeshOutput, MAX_VERTICES> vertices, 
+	out Vertices<FragmentParameter, MAX_VERTICES> vertices, 
 	out Indices<uint3, MAX_PRIMITIVES> indices
 ){
 	InstanceData inst = pScene.instances[meshPayload.instanceId[groupID]];
 	MeshData md = pScene.meshData[meshPayload.instanceId[groupID]];
 	MeshletDescription m = pScene.meshletInfos[meshPayload.meshletId[groupID]];
-	const uint vertexLoops = (MAX_VERTICES + MESH_GROUP_SIZE - 1) / MESH_GROUP_SIZE;
+	SetMeshOutputCounts(m.vertexCount, m.primitiveCount);
-	for(uint i = threadID; i < MAX_VERTICES; i += MESH_GROUP_SIZE)
+	
+	uint p = min(threadID, m.primitiveCount - 1);
+	{
+		uint local_idx0 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 0];
+		uint local_idx1 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 1];
+		uint local_idx2 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 2];
+		indices[p] = uint3(local_idx0, local_idx1, local_idx2);
+	}
+	p = min(threadID + 32, m.primitiveCount - 1);
+	{
+		uint local_idx0 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 0];
+		uint local_idx1 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 1];
+		uint local_idx2 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 2];
+		indices[p] = uint3(local_idx0, local_idx1, local_idx2);
+	}
+	p = min(threadID + 64, m.primitiveCount - 1);
+	{
+		uint local_idx0 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 0];
+		uint local_idx1 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 1];
+		uint local_idx2 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 2];
+		indices[p] = uint3(local_idx0, local_idx1, local_idx2);
+	}
+	p = min(threadID + 96, m.primitiveCount - 1);
+	{
+		uint local_idx0 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 0];
+		uint local_idx1 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 1];
+		uint local_idx2 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 2];
+		indices[p] = uint3(local_idx0, local_idx1, local_idx2);
+	}
+	
+	GroupMemoryBarrierWithGroupSync();
+	for(uint i = threadID; i < MAX_VERTICES; i+=MESH_GROUP_SIZE)
 	{
 		uint v = min(i, m.vertexCount - 1);
-		InterlockedMax(gs_numVertices, v + 1);
 		{
 			int vertexIndex = pScene.vertexIndices[m.vertexOffset + v];
-			gs_vertices[v] = pVertexData.getAttributes(md.indicesOffset + vertexIndex, inst.transformMatrix);
+			vertices[v] = pVertexData.getAttributes(md.indicesOffset + vertexIndex).getParameter(inst.transformMatrix);
 		}
 	}
-	GroupMemoryBarrierWithGroupSync();
-	const uint primitiveLoops = (MAX_PRIMITIVES + MESH_GROUP_SIZE - 1) / MESH_GROUP_SIZE;
-	for(uint i = threadID; i < MAX_PRIMITIVES; i += MESH_GROUP_SIZE)
-	{
-		uint p = min(i, m.primitiveCount - 1);
-		InterlockedMax(gs_numPrimitives, p + 1);
-		{
-			uint32_t local_idx0 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 0];
-			uint32_t local_idx1 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 1];
-			uint32_t local_idx2 = pScene.primitiveIndices[m.primitiveOffset + (p * 3) + 2];
-			gs_indices[p].x = local_idx0;
-			gs_indices[p].y = local_idx1;
-			gs_indices[p].z = local_idx2;
-		}
-	}
-	GroupMemoryBarrierWithGroupSync();
-	SetMeshOutputCounts(gs_numVertices, gs_numPrimitives);
-	GroupMemoryBarrierWithGroupSync();
-	
-	uint v = threadID;
-	v = min(v, m.vertexCount - 1);
-	FragmentParameter parameter = gs_vertices[v].getParameter(inst.transformMatrix);
-	vertices[v].parameter = parameter;
-	if(vertexLoops >= 1)
-	{
-		uint v = threadID + MESH_GROUP_SIZE;
-		v = min(v, m.vertexCount - 1);
-		FragmentParameter parameter = gs_vertices[v].getParameter(inst.transformMatrix);
-		vertices[v].parameter = parameter;
-	}
-
-	uint p = threadID;
-	p = min(p, m.primitiveCount - 1);
-	indices[p] = gs_indices[p];
-	
-	if(primitiveLoops >= 1)
-	{
-		p += MESH_GROUP_SIZE;
-		p = min(p, m.primitiveCount - 1);
-		indices[p] = gs_indices[p];
-	}
-	if(primitiveLoops >= 2)
-	{
-		p += MESH_GROUP_SIZE;
-		p = min(p, m.primitiveCount - 1);
-		indices[p] = gs_indices[p];
-	}
-	if(primitiveLoops >= 3)
-	{
-		p += MESH_GROUP_SIZE;
-		p = min(p, m.primitiveCount - 1);
-		indices[p] = gs_indices[p];
-	}
 }

res/shaders/lib/MaterialParameter.slang

@@ -51,28 +51,31 @@ struct FragmentParameter
 // data retrieved from VertexData
 struct VertexAttributes
 {
+	float3 position_MS;
 	float3 normal_MS;
 	float3 tangent_MS;
 	float3 biTangent_MS;
-	float3 position_WS;
-	float4 position_CS;
 	float2 texCoords;
 	float3 vertexColor;
 	FragmentParameter getParameter(float4x4 transformMatrix)
 	{
+		float4 modelPos = float4(position_MS, 1);
+		float4 worldPos = mul(transformMatrix, modelPos);
+		float4 viewPos = mul(pViewParams.viewMatrix, worldPos);
+		float4 clipPos = mul(pViewParams.projectionMatrix, viewPos);
 		float3 tangent_WS = mul(transformMatrix, float4(normalize(tangent_MS), 0)).xyz;
 		float3 biTangent_WS = mul(transformMatrix, float4(normalize(biTangent_MS), 0)).xyz;
 		float3 normal_WS = mul(transformMatrix, float4(normalize(normal_MS), 0)).xyz;
 		// Transforms from world space into tangent space
 		float3x3 tbn = transpose(float3x3(tangent_WS, biTangent_WS, normal_WS));
 		FragmentParameter result;
-		result.position_TS = mul(tbn, position_WS);
+		result.position_TS = mul(tbn, worldPos.xyz);
-		result.viewDir_TS = mul(tbn, pViewParams.cameraPos_WS.xyz - position_WS);
+		result.viewDir_TS = mul(tbn, pViewParams.cameraPos_WS.xyz - worldPos.xyz);
 		result.normal_WS = normal_WS;
 		result.tangent_WS = tangent_WS;
 		result.biTangent_WS = biTangent_WS;
-		result.position_WS = position_WS;
+		result.position_WS = worldPos.xyz;
-		result.position_CS = position_CS;
+		result.position_CS = clipPos;
 		result.texCoords = texCoords;
 		result.vertexColor = vertexColor;
 		return result;

res/shaders/lib/StaticMeshVertexData.slang

@@ -4,18 +4,13 @@ import MaterialParameter;
 
 struct StaticMeshVertexData : IVertexData
 {
-	VertexAttributes getAttributes(uint index, float4x4 transform)
+	VertexAttributes getAttributes(uint index)
 	{
 		VertexAttributes attributes;
-		float4 modelPos = float4(positions[3 * index + 0], positions[3 * index + 1], positions[3 * index + 2], 1);
+		attributes.position_MS = float3(positions[3 * index + 0], positions[3 * index + 1], positions[3 * index + 2]);
-		float4 worldPos = mul(transform, modelPos);
-		float4 viewPos = mul(pViewParams.viewMatrix, worldPos);
-		float4 clipPos = mul(pViewParams.projectionMatrix, viewPos);
 		attributes.normal_MS = float3(normals[3 * index + 0], normals[3 * index + 1], normals[3 * index + 2]);
 		attributes.tangent_MS = float3(tangents[3 * index + 0], tangents[3 * index + 1], tangents[3 * index + 2]);
 		attributes.biTangent_MS = float3(biTangents[3 * index + 0], biTangents[3 * index + 1], biTangents[3 * index + 2]);
-		attributes.position_WS = worldPos.xyz;
-		attributes.position_CS = clipPos;
 		attributes.texCoords = float2(texCoords[2 * index + 0], texCoords[2 * index + 1]);
 		attributes.vertexColor = float3(color[3 * index + 0], color[3 * index + 1], color[3 * index + 2]);
 		return attributes;

res/shaders/lib/VertexData.slang

@@ -2,7 +2,7 @@ import MaterialParameter;
 
 interface IVertexData
 {
-	VertexAttributes getAttributes(uint index, float4x4 transform);
+	VertexAttributes getAttributes(uint index);
 };
 layout(set = 1)
 ParameterBlock<IVertexData> pVertexData;

src/Engine/Graphics/Vulkan/Graphics.cpp

@@ -399,7 +399,7 @@ void Graphics::pickPhysicalDevice()
         {
             if (std::strcmp(VK_EXT_MESH_SHADER_EXTENSION_NAME, extensionProps[i].extensionName) == 0)
             {
-                //meshShadingEnabled = true;
+                meshShadingEnabled = true;
                 break;
             }
         }

src/Engine/Graphics/Vulkan/Shader.cpp

@@ -55,8 +55,12 @@ void Shader::create(const ShaderCreateInfo& createInfo)
     slang::TargetDesc vulkan;
     vulkan.profile = globalSession->findProfile("sm_6_6");
     vulkan.format = SLANG_SPIRV;
-    sessionDesc.targetCount = 1;
+    slang::TargetDesc glsl;
-    sessionDesc.targets = &vulkan;
+    glsl.profile = globalSession->findProfile("sm_6_6");
+    glsl.format = SLANG_GLSL;
+    slang::TargetDesc targets[2] = { vulkan, glsl };
+    sessionDesc.targetCount = 2;
+    sessionDesc.targets = targets;
     StaticArray<const char*, 3> searchPaths = {"shaders/", "shaders/lib/", "shaders/generated/"};
     sessionDesc.searchPaths = searchPaths.data();
     sessionDesc.searchPathCount = searchPaths.size();
@@ -132,4 +136,15 @@ void Shader::create(const ShaderCreateInfo& createInfo)
 
     hash = CRC::Calculate(entryPointName.data(), entryPointName.size(), CRC::CRC_32());
     hash = CRC::Calculate(kernelBlob->getBufferPointer(), kernelBlob->getBufferSize(), CRC::CRC_32(), hash);
+
+    specializedComponent->getEntryPointCode(
+        0,
+        1,
+        kernelBlob.writeRef(),
+        diagnostics.writeRef()
+    );
+    CHECK_DIAGNOSTICS();
+    std::ofstream shaderStream(createInfo.name + createInfo.entryPoint + ".glsl");
+    shaderStream << (char*)kernelBlob->getBufferPointer();
+    shaderStream.close();
 }