// Copyright (c) 2018-2020 NVIDIA Corporation // // SPDX-License-Identifier: CC-BY-4.0 [[mesh]] = Mesh Shading <> and <> operate in workgroups to produce a collection of primitives that will be processed by subsequent stages of the graphics pipeline. Work on the mesh pipeline is initiated by the application <> a set of mesh tasks organized in global workgroups. If the optional task shader is active, each workgroup triggers the execution of task shader invocations that will create a new set of mesh workgroups upon completion. Each of these created workgroups, or each of the original workgroups if no task shader is present, triggers the execution of mesh shader invocations. Each mesh shader workgroup emits zero or more output primitives along with the group of vertices and their associated data required for each output primitive. [[mesh-task-input]] == Task Shader Input For every workgroup issued via the drawing commands a group of task shader invocations is executed. There are no inputs other than the builtin workgroup identifiers. [[mesh-task-output]] == Task Shader Output The task shader can emit zero or more mesh workgroups to be generated. ifdef::VK_NV_mesh_shader[] Shaders using the code:TaskNV {ExecutionModel} can do so using the <> code:TaskCountNV. This value must: be less than or equal to sname:VkPhysicalDeviceMeshShaderPropertiesNV::pname:maxTaskOutputCount. endif::VK_NV_mesh_shader[] ifdef::VK_EXT_mesh_shader[] Shaders using the code:TaskEXT {ExecutionModel} can do so using the code:OpEmitMeshTasksEXT instruction. The code:groupCountX, code:groupCountY and code:groupCountZ arguments passed to this instruction must: be less than or equal to the respective dimension within sname:VkPhysicalDeviceMeshShaderPropertiesEXT::pname:maxMeshWorkGroupCount. The product of these arguments must: be less than or equal to sname:VkPhysicalDeviceMeshShaderPropertiesEXT::pname:maxMeshWorkGroupTotalCount. endif::VK_EXT_mesh_shader[] The task shader can also pass user-defined data to all mesh shader invocations that it creates. ifdef::VK_NV_mesh_shader[] Shaders using the code:TaskNV {ExecutionModel} can do so by writing to output variables that are decorated with code:PerTaskNV. They are available as inputs in mesh shaders. endif::VK_NV_mesh_shader[] ifdef::VK_EXT_mesh_shader[] Shaders using the code:TaskEXT {ExecutionModel} can do so by writing to a payload variable with code:TaskPayloadWorkgroupEXT storage class that is passed to the code:OpEmitMeshTasksEXT instruction. endif::VK_EXT_mesh_shader[] [[mesh-generation]] == Mesh Generation If a task shader exists, the mesh assembler creates a variable amount of mesh workgroups depending on each task's output. If there is no task shader, the drawing commands emit the mesh shader invocations directly. [[mesh-input]] == Mesh Shader Input The only inputs available to the mesh shader are variables identifying the specific workgroup and invocation and, if applicable, ifdef::VK_NV_mesh_shader[] any outputs written as code:PerTaskNV endif::VK_NV_mesh_shader[] ifdef::VK_NV_mesh_shader+VK_EXT_mesh_shader[or] ifdef::VK_EXT_mesh_shader[] the payload variable passed to the code:OpEmitMeshTasksEXT instruction endif::VK_EXT_mesh_shader[] by the task shader that spawned the mesh shader's workgroup. The mesh shader can operate without a task shader as well. [[mesh-output]] == Mesh Shader Output A mesh shader generates primitives in one of three output modes: points, lines, or triangles. ifdef::VK_NV_mesh_shader[] For shaders using the code:MeshNV {ExecutionModel} the primitive mode is specified in the shader using an code:OpExecutionMode instruction with the code:OutputPoints, code:OutputLinesNV, or code:OutputTrianglesNV modes, respectively. endif::VK_NV_mesh_shader[] ifdef::VK_EXT_mesh_shader[] For shaders using the code:MeshEXT {ExecutionModel} the primitive mode is specified in the shader using an code:OpExecutionMode instruction with the code:OutputPoints, code:OutputLinesEXT, or code:OutputTrianglesEXT modes, respectively. endif::VK_EXT_mesh_shader[] Each mesh shader must: include exactly one output primitive mode. ifdef::VK_NV_mesh_shader[] For shaders using the code:MeshNV {ExecutionModel} the maximum output vertex count is specified as a literal in the shader using an code:OpExecutionMode instruction with the mode set to code:OutputVertices and must: be less than or equal to sname:VkPhysicalDeviceMeshShaderPropertiesNV::pname:maxMeshOutputVertices. endif::VK_NV_mesh_shader[] ifdef::VK_EXT_mesh_shader[] For shaders using the code:MeshEXT {ExecutionModel} the maximum output vertex count is specified as a literal in the shader using an code:OpExecutionMode instruction with the mode set to code:OutputVertices and must: be less than or equal to sname:VkPhysicalDeviceMeshShaderPropertiesEXT::pname:maxMeshOutputVertices. endif::VK_EXT_mesh_shader[] ifdef::VK_NV_mesh_shader[] For shaders using the code:MeshNV {ExecutionModel} the maximum output primitive count is specified as a literal in the shader using an code:OpExecutionMode instruction with the mode set to code:OutputPrimitivesNV and must: be less than or equal to sname:VkPhysicalDeviceMeshShaderPropertiesNV::pname:maxMeshOutputPrimitives. endif::VK_NV_mesh_shader[] ifdef::VK_EXT_mesh_shader[] For shaders using the code:MeshEXT {ExecutionModel} the maximum output primitive count is specified as a literal in the shader using an code:OpExecutionMode instruction with the mode set to code:OutputPrimitivesEXT, and must: be less than or equal to sname:VkPhysicalDeviceMeshShaderPropertiesEXT::pname:maxMeshOutputPrimitives. endif::VK_EXT_mesh_shader[] ifdef::VK_NV_mesh_shader[] For shaders using the code:MeshNV {ExecutionModel} the number of primitives output by the mesh shader is provided via writing to the <> code:PrimitiveCountNV and must: be less than or equal to the maximum output primitive count specified in the shader. A variable decorated with code:PrimitiveIndicesNV is an output array of local index values into the vertex output arrays from which primitives are assembled according to the output primitive type. endif::VK_NV_mesh_shader[] ifdef::VK_EXT_mesh_shader[] For shaders using the code:MeshEXT {ExecutionModel} the number of vertices and primitives output by the mesh shader is provided via calling the code:OpSetMeshOutputsEXT instruction. The code:vertexCount argument must: be less than or equal to the maximum output vertex count specified in the shader. The code:primitiveCount argument must: be less than or equal to the maximum output primitive count specified in the shader. Depending on the output primitive mode an appropriately-decorated variable is the output array of local index values into the vertex output arrays from which primitives are assembled according to the output primitive type: * code:OutputPoints uses the code:PrimitivePointIndicesEXT decoration. * code:OutputLinesEXT uses the code:PrimitiveLineIndicesEXT decoration. * code:OutputTrianglesEXT uses the code:PrimitiveTriangleIndicesEXT decoration. endif::VK_EXT_mesh_shader[] These resulting primitives are then further processed as described in <>. ifdef::VK_EXT_mesh_shader[] With the exception of primitive indices, all output built-ins and custom attributes count towards the total storage size occupied by output variables in mesh shaders. This size can be calculated as follows, taking into account the fact that the number of effective scalar attributes is 4 times the number of effective locations used according to the <>. Let latexmath:[v] be the number of views, latexmath:[n_v] be the number of effective scalar per-vertex attributes not dependent on code:ViewIndex, latexmath:[n_{vpv}] be the number of effective scalar per-vertex attributes dependent on code:ViewIndex, latexmath:[m_v] be the maximum number of vertices specified by the code:OutputVertices {ExecutionMode}, latexmath:[g_v] be pname:meshOutputPerVertexGranularity, latexmath:[n_p] be the number of effective scalar per-primitive attributes not dependent on code:ViewIndex, latexmath:[n_{ppv}] be the number of effective scalar per-primitive attributes dependent on code:ViewIndex, latexmath:[m_p] be the maximum number of primitives specified by the code:OutputPrimitivesEXT {ExecutionMode} and latexmath:[g_p] be pname:meshOutputPerPrimitiveGranularity: [latexmath] ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ (n_v + (n_{vpv} \times v)) \times 4 \times \mathrm{align}(m_v, g_v) + (n_p + (n_{ppv} \times v)) \times 4 \times \mathrm{align}(m_p, g_p) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ endif::VK_EXT_mesh_shader[] ifdef::VK_NV_mesh_shader[] [[mesh-output-perview]] == Mesh Shader Per-View Outputs The mesh shader outputs decorated with the code:PositionPerViewNV, code:ClipDistancePerViewNV, code:CullDistancePerViewNV, code:LayerPerViewNV, and code:ViewportMaskPerViewNV built-in decorations are the per-view versions of the single-view variables with equivalent names (that is code:Position, code:ClipDistance, code:CullDistance, code:Layer, and code:ViewportMaskNV, respectively). If a shader statically assigns a value to any element of a per-view array it must: not statically assign a value to the equivalent single-view variable. Each of these outputs is considered arrayed, with separate values for each view. The view number is used to index the first dimension of these arrays. The second dimension of the code:ClipDistancePerViewNV, and code:CullDistancePerViewNV arrays have the same requirements as the code:ClipDistance, and code:CullDistance arrays. If a mesh shader output is _per-view_, the corresponding fragment shader input is taken from the element of the per-view output array that corresponds to the view that is currently being processed by the fragment shader. ifdef::VK_EXT_mesh_shader[] These _per-view_ outputs are available only in shaders using the code:MeshNV {ExecutionModel}. They are not available in shaders using the code:MeshEXT {ExecutionModel}. endif::VK_EXT_mesh_shader[] endif::VK_NV_mesh_shader[] [[mesh-ordering]] == Mesh Shader Primitive Ordering Following guarantees are provided for the relative ordering of primitives produced by a mesh shader, as they pertain to <>. * When a task shader is used, mesh workgroups spawned from lower tasks will be ordered prior those workgroups from subsequent tasks. * All output primitives generated from a given mesh workgroup are passed to subsequent pipeline stages before any output primitives generated from subsequent input workgroups. * All output primitives within a mesh workgroup, will be generated in the ordering provided by the builtin primitive indexbuffer (from low address to high address).