大网数据广告活动
Web 标准与前端开发

GPUDevice: createRenderPipeline() method

· 52CE 我爱测

The **`createRenderPipeline()`** method of the GPUDevice interface creates a GPURenderPipeline that can control the vertex and fragment shader stages and be used in a GPURenderPassEncoder or GPURenderBundleEncoder.

WebGPU API

The createRenderPipeline() method of the GPUDevice interface creates a GPURenderPipeline that can control the vertex and fragment shader stages and be used in a GPURenderPassEncoder or GPURenderBundleEncoder.

Syntax

createRenderPipeline(descriptor)

Parameters

  • descriptor
    • : An object containing the following properties:
      • depthStencil
      • fragment
        • : An object (see fragment object structure) describing the fragment shader entry point of the pipeline and its output colors. If no fragment shader entry point is defined, the pipeline will not produce any color attachment outputs, but it still performs rasterization and produces depth values based on the vertex position output. Depth testing and stencil operations can still be used.
      • label
        • : A string providing a label that can be used to identify the object, for example in GPUError messages or console warnings.
      • layout
        • : Defines the layout (structure, purpose, and type) of all the GPU resources (buffers, textures, etc.) used during the execution of the pipeline. Possible values are:
          • A GPUPipelineLayout object, created using GPUDevice.createPipelineLayout(), which allows the GPU to figure out how to run the pipeline most efficiently ahead of time.
          • A string of "auto", which causes the pipeline to generate an implicit bind group layout based on any bindings defined in the shader code. If "auto" is used, the generated bind group layouts may only be used with the current pipeline.
      • multisample
      • primitive
        • : An object (see primitive object structure) describing how a pipeline constructs and rasterizes primitives from its vertex inputs.
      • vertex
        • : An object (see vertex object structure) describing the vertex shader entry point of the pipeline and its input buffer layouts.

depthStencil object structure

The depthStencil object can contain the following properties:

  • depthBias

    • : A number representing a constant depth bias that is added to each fragment. If omitted, depthBias defaults to 0.

      [!NOTE] The depthBias, depthBiasClamp, and depthBiasSlopeScale properties must be set to 0 for line and point topologies, i.e., if topology is set to "line-list", "line-strip", or "point-list". If not, a GPUValidationError will be generated and the returned GPURenderPipeline will be invalid.

  • depthBiasClamp

    • : A number representing the maximum depth bias of a fragment. If omitted, depthBiasClamp defaults to 0.
  • depthBiasSlopeScale

    • : A number representing a depth bias that scales with the fragment's slope. If omitted, depthBiasSlopeScale defaults to 0.
  • depthCompare

    • : An enumerated value specifying the comparison operation used to test fragment depths against depthStencilAttachment depth values. Possible values are:

      • "never": Comparison tests never pass.
      • "less": A provided value passes the comparison test if it is less than the sampled value.
      • "equal": A provided value passes the comparison test if it is equal to the sampled value.
      • "less-equal": A provided value passes the comparison test if it is less than or equal to the sampled value.
      • "greater": A provided value passes the comparison test if it is greater than the sampled value.
      • "not-equal": A provided value passes the comparison test if it is not equal to the sampled value.
      • "greater-equal": A provided value passes the comparison test if it is greater than or equal to the sampled value.
      • "always": Comparison tests always pass.

      depthCompare is not required if the specified format does not have a depth component, or if the comparison operation is not used.

  • depthWriteEnabled

    • : A boolean. A value of true specifies that the GPURenderPipeline can modify depthStencilAttachment depth values after creation. Setting it to false means it cannot.

      depthWriteEnabled is not required if the specified format does not have a depth component.

  • format

    • : An enumerated value specifying the depthStencilAttachment format that the GPURenderPipeline will be compatible with. See the specification's Texture Formats section for all the available format values.
  • stencilBack

    • : An object that defines how stencil comparisons and operations are performed for back-facing primitives. Its properties can include:
      • compare

        • : An enumerated value specifying the comparison operation used when testing fragments against depthStencilAttachment stencil values. Possible values are the same as for the depthCompare property; see above. If omitted, compare defaults to "always".
      • depthFailOp

        • : An enumerated value specifying the stencil operation performed if the fragment depth comparison described by depthCompare fails. Possible values are:

          • "decrement-clamp": Decrement the current render state stencil value, clamping it to 0.
          • "decrement-wrap": Decrement the current render state stencil value, wrapping it to the maximum representable value of the depthStencilAttachment's stencil aspect if the value goes below 0.
          • "invert": Bitwise-invert the current render state stencil value.
          • "increment-clamp": Increments the current render state stencil value, clamping it to the maximum representable value of the depthStencilAttachment's stencil aspect.
          • "increment-wrap": Increments the current render state stencil value, wrapping it to zero if the value exceeds the maximum representable value of the depthStencilAttachment's stencil aspect.
          • "keep": Keep the current stencil value.
          • "replace": Set the stencil value to the current render state stencil value.
          • "zero": Set the stencil value to 0.

          If omitted, depthFailOp defaults to "keep".

          [!NOTE] The render state stencil value is initialized to 0 at the start of a render pass.

      • failOp

        • : An enumerated value specifying the stencil operation performed if the fragment stencil comparison test described by compare fails. Possible and default values are the same as for depthFailOp.
      • passOp

        • : An enumerated value specifying the stencil operation performed if the fragment stencil comparison test described by compare passes. Possible and default values are the same as for depthFailOp.
  • stencilFront

    • : An object that defines how stencil comparisons and operations are performed for front-facing primitives. Its properties are the same as for stencilBack.
  • stencilReadMask

    • : A bitmask controlling which depthStencilAttachment stencil value bits are read when performing stencil comparison tests. If omitted, stencilReadMask defaults to 0xFFFFFFFF.
  • stencilWriteMask

    • : A bitmask controlling which depthStencilAttachment stencil value bits are written to when performing stencil operations. If omitted, stencilWriteMask defaults to 0xFFFFFFFF.

[!NOTE] depthStencilAttachment values are specified during GPUCommandEncoder.beginRenderPass() calls, when the GPURenderPipeline is actually used to perform a render pass.

fragment object structure

The fragment object contains an array of objects, each of which can contain the following properties:

  • constants

    • : A sequence of record types, with the structure (id, value), representing override values for WGSL constants that can be overridden in the pipeline. These behave like ordered maps. In each case, the id is a key used to identify or select the record, and the constant is an enumerated value representing a WGSL.

      Depending on which constant you want to override, the id may take the form of the numeric ID of the constant, if one is specified, or otherwise the constant's identifier name.

      A code snippet providing override values for several overridable constants might look like this:

      ({
        // …
        constants: {
          0: false,
          1200: 3.0,
          1300: 2.0,
          width: 20,
          depth: -1,
          height: 15,
        },
      });
      
  • entryPoint

    • : The name of the function in the module that this stage will use to perform its work. The corresponding shader function must have the @fragment attribute to be identified as this entry point. See Entry Point Declaration for more information.

      You can omit the entryPoint property if your shader code contains a single function with the @fragment attribute set — the browser will use this as the default entry point. If entryPoint is omitted and the browser cannot determine a default entry point, a GPUValidationError is generated and the resulting GPURenderPipeline will be invalid.

  • module

    • : A GPUShaderModule object containing the WGSL code that this programmable stage will execute.
  • targets

    • : an array of objects representing color states that represent configuration details for the colors output by the fragment shader stage. These objects can include the following properties:
      • blend

        • : An object that describes a blend mode to be applied to the output color. blend has two properties:

          • alpha
            • : Describes the alpha channel value.
          • color
            • : Describes the color value.

          alpha and color both take an object as a value that can include the following properties:

          • dstFactor

            • : An enumerated value that defines the blend factor operation to be performed on values from the target attachment. Possible values are:

              • "constant"
              • "dst"
              • "dst-alpha"
              • "one"
              • "one-minus-dst"
              • "one-minus-src"
              • "one-minus-src1"
              • "one-minus-src-alpha"
              • "one-minus-src1-alpha"
              • "one-minus-dst-alpha"
              • "one-minus-constant"
              • "src"
              • "src1"
              • "src-alpha"
              • "src1-alpha"
              • "src-alpha-saturated"
              • "zero"

              If omitted, dstFactor defaults to "zero".

              [!NOTE] The dual-source-blending feature needs to be enabled for the src1, one-minus-src1, src1-alpha, and one-minus-src1-alpha blend factor operations to be used successfully. If not, a GPUValidationError is generated.

          • operation

            • : An enumerated value that defines the algorithm used to combine source and destination blend factors, to calculate the final values written to the target attachment components. Possible values are:

              • "add"
              • "max"
              • "min"
              • "reverse-subtract"
              • "subtract"

              If omitted, operation defaults to "add".

          • srcFactor

            • : An enumerated value that defines the blend factor operation to be performed on values from the fragment shader. Possible values are the same as for dstFactor. If omitted, srcFactor defaults to "one".

          [!NOTE] For a detailed explanation of the algorithms defined by each dstFactor/srcFactor and operation enumerated value, see the Blend State section of the specification.

      • format

        • : An enumerated value specifying the required format for output colors. See the specification's Texture Formats section for all the available format values.

          [!NOTE] For the r32float, rg32float, and rgba32float formats to be used with blending, the float32-blendable feature must be available in the device.

      • writeMask

        • : One or more bitwise flags defining the write mask to apply to the color target state. Possible flag values are:

          • GPUColorWrite.RED
          • GPUColorWrite.GREEN
          • GPUColorWrite.BLUE
          • GPUColorWrite.ALPHA
          • GPUColorWrite.ALL

          If omitted, writeMask defaults to GPUColorWrite.ALL.

          Note that multiple flags can be specified by separating values with bitwise OR, for example: GPUColorWrite.RED | GPUColorWrite.ALPHA.

multisample object structure

The multisample object can contain the following properties:

  • alphaToCoverageEnabled

    • : A boolean. A value of true indicates that a fragment's alpha channel should be used to generate a sample coverage mask. If omitted, alphaToCoverageEnabled defaults to false.
  • count

    • : A number that defines the number of samples per pixel. The pipeline will be compatible only with attachment textures (colorAttachments and depthStencilAttachments) with matching sampleCounts (see GPUTexture).

      If omitted, count defaults to 1.

  • mask

    • : A bitmask that determines which samples are written to. If omitted, mask defaults to 0xFFFFFFFF.

[!NOTE] colorAttachment and depthStencilAttachment values are specified during GPUCommandEncoder.beginRenderPass() calls, when the GPURenderPipeline is actually used to perform a render pass.

primitive object structure

The primitive object can contain the following properties:

  • cullMode

    • : An enumerated value that defines which polygon orientation will be culled, if any. Possible values are:

      • "back": Back-facing polygons are culled.
      • "front": Front-facing polygons are culled.
      • "none": No polygons are culled.

      If omitted, cullMode defaults to "none".

  • frontFace

    • : An enumerated value that defines which polygons are considered front-facing. Possible values are:

      • "ccw": Polygons with vertices whose framebuffer coordinates are given in counter-clockwise order.
      • "cw": Polygons with vertices whose framebuffer coordinates are given in clockwise order.

      If omitted, frontFace defaults to "ccw".

      [!NOTE] The frontFace and cullMode values have no effect on the "point-list", "line-list", or "line-strip" topologies.

  • stripIndexFormat

    • : An enumerated value that determines the index buffer format and primitive restart value in the case of pipelines with strip topologies ("line-strip" or "triangle-strip"). The primitive restart value specifies which index value indicates that a new primitive should be started rather than continuing to construct the strip with the prior indexed vertices. Possible values are:

      • "uint16": Indicates a byte size of 2 and a primitive restart value of 0xFFFF.
      • "uint32": Indicates a byte size of 4 and a primitive restart value of 0xFFFFFFFF.

      GPU primitive states that specify a strip primitive topology must specify a strip index format if they are used for indexed draws (for example, via GPURenderPassEncoder.drawIndexed()) so that the primitive restart value that will be used is known at pipeline creation time. Pipelines with list primitive topologies ("line-list", "point-list", or "triangle-list") should not specify a stripIndexFormat value. They will instead use the index format passed to, for example, GPURenderPassEncoder.setIndexBuffer() when doing indexed rendering.

  • topology

    • : An enumerated value that defines the type of primitive to be constructed from the specified vertex inputs. Possible values are:

      • "line-list": Each consecutive pair of two vertices defines a line primitive.
      • "line-strip": Each vertex after the first defines a line primitive between it and the previous vertex.
      • "point-list": Each vertex defines a point primitive.
      • "triangle-list": Each consecutive triplet of three vertices defines a triangle primitive.
      • "triangle-strip": Each vertex after the first two defines a triangle primitive between it and the previous two vertices.

      If omitted, topology defaults to "triangle-list".

  • unclippedDepth

    • : A boolean. A value of true indicates that depth clipping is disabled. If omitted, unclippedDepth defaults to false. Note that to control depth clipping, the depth-clip-control GPUSupportedFeatures must be enabled in the GPUDevice.

      [!NOTE] The depth-clip-control feature needs to be enabled for the unclippedDepth property to be used successfully. If not, a GPUValidationError is generated.

vertex object structure

The vertex object can contain the following properties:

  • constants

    • : A sequence of record types, with the structure (id, value), representing override values for WGSL constants that can be overridden in the pipeline. These behave like ordered maps. In each case, the id is a key used to identify or select the record, and the constant is an enumerated value representing a WGSL.

      Depending on which constant you want to override, the id may take the form of the numeric ID of the constant, if one is specified, or otherwise the constant's identifier name.

      A code snippet providing override values for several overridable constants might look like this:

      ({
        // …
        constants: {
          0: false,
          1200: 3.0,
          1300: 2.0,
          width: 20,
          depth: -1,
          height: 15,
        },
      });
      
  • entryPoint

    • : The name of the function in the module that this stage will use to perform its work. The corresponding shader function must have the @vertex attribute to be identified as this entry point. See Entry Point Declaration for more information.

      You can omit the entryPoint property if your shader code contains a single function with the @vertex attribute set — the browser will use this as the default entry point. If entryPoint is omitted and the browser cannot determine a default entry point, a GPUValidationError is generated and the resulting GPURenderPipeline will be invalid.

  • module

    • : A GPUShaderModule object containing the WGSL code that this programmable stage will execute.
  • buffers

    • : An array of objects, each representing the expected layout of a vertex buffer used in the pipeline. Each object can contain the following properties:
      • arrayStride
        • : A number representing the stride, in bytes, between the different structures (e.g., vertices) inside the buffer.
      • attributes
        • : An array of objects defining the layout of the vertex attributes within each structure. Each object has the following properties:
          • format
            • : An enumerated value that specifies the format of the vertex. For all the available values, see the GPUVertexFormat definition in the specification.
          • offset
            • : A number specifying the offset, in bytes, from the beginning of the structure to the data for the attribute.
          • shaderLocation
            • : The numeric location associated with this attribute, which will correspond with a @location attribute declared in the WGSL code of the associated GPUShaderModule referenced in the vertex object's module property.
      • stepMode
        • : An enumerated value that defines whether the separate structures inside the buffer represent vertices or instances. Possible values are:

          • "instance": Each structure is an instance — the address is advanced by arrayStride for each instance.
          • "vertex": Each structure is a vertex — the address is advanced by arrayStride for each vertex, and reset between instances.

          If omitted, stepMode defaults to "vertex".

Return value

A GPURenderPipeline object instance.

Validation

The following criteria must be met when calling createRenderPipeline(), otherwise a GPUValidationError is generated and an invalid GPURenderPipeline object is returned:

  • For depthStencil objects:
    • format is a depth-or-stencil format.
    • The depthBias, depthBiasClamp, and depthBiasSlopeScale properties are set to <code>0</code> for line and point topologies, i.e., if topology is set to "line-list", "line-strip", or "point-list".
    • If depthWriteEnabled is true or depthCompare is not "always", format has a depth component.
    • If stencilFront or stencilBack's properties are not at their default values, format has a stencil component.
  • For fragment objects:
    • targets.length is less than or equal to the GPUDevice's maxColorAttachments GPUSupportedLimits.
    • For each target, writeMask's numeric equivalent is less than 16.
    • If any of the used blend factor operations use the source alpha channel (for example "src-alpha-saturated"), the output has an alpha channel (that is, it must be a vec4).
    • If the src1, one-minus-src1, src1-alpha, or one-minus-src1-alpha blend factor operations are used, the dual-source-blending feature is enabled.
    • If the entryPoint property is omitted, the shader code contains a single fragment shader entry point function for the browser to use as the default entry point.
  • For primitive objects:
    • If the unclippedDepth property is used, the depth-clip-control feature is enabled.
  • For vertex objects:
    • If the entryPoint property is omitted, the shader code contains a single vertex shader entry point function for the browser to use as the default entry point.

Examples

[!NOTE] The WebGPU samples feature many more examples.

Basic example

Our basic render demo provides an example of the construction of a valid render pipeline descriptor object, which is then used to create a GPURenderPipeline via a createRenderPipeline() call.

// …

const vertexBuffers = [
  {
    attributes: [
      {
        shaderLocation: 0, // position
        offset: 0,
        format: "float32x4",
      },
      {
        shaderLocation: 1, // color
        offset: 16,
        format: "float32x4",
      },
    ],
    arrayStride: 32,
    stepMode: "vertex",
  },
];

const pipelineDescriptor = {
  vertex: {
    module: shaderModule,
    entryPoint: "vertex_main",
    buffers: vertexBuffers,
  },
  fragment: {
    module: shaderModule,
    entryPoint: "fragment_main",
    targets: [
      {
        format: navigator.gpu.getPreferredCanvasFormat(),
      },
    ],
  },
  primitive: {
    topology: "triangle-list",
  },
  layout: "auto",
};

const renderPipeline = device.createRenderPipeline(pipelineDescriptor);

// …

Specifications

Browser compatibility

See also

来源:MDN Web Docs / Mozilla 贡献者 · 作者与贡献记录 · CC BY-SA 2.5。离线整理了版式、链接和媒体显示。

资料来源与适用范围

资料来源
MDN Web Docs / Mozilla 贡献者
适用范围
请按文中版本与环境说明核对
内容语言
中文;命令、路径和错误标识保留技术原文
本地收录日期
2026-11-15
授权说明
CC BY-SA 2.5

正文与所需资源由本站本地存储。查看来源及更新记录 ↗

请按适用版本和实际日志确认原因;本站没有对每一种系统或设备逐项执行文中的操作。