WebGPU

1. Introduction

This section is non-normative.

Graphics Processing Units, or GPUs for short, have been essential in enabling rich rendering and computational applications in personal computing. WebGPU is an API that exposes the capabilities of GPU hardware for the Web. The API is designed from the ground up to efficiently map to the Vulkan, Direct3D 12, and Metal native GPU APIs. WebGPU is not related to WebGL and does not explicitly target OpenGL ES.

WebGPU sees physical GPU hardware as GPUAdapters. It provides a connection to an adapter via GPUDevice, which manages resources, and the device’s GPUQueues, which execute commands. GPUDevice may have its own memory with high-speed access to the processing units. GPUBuffer and GPUTexture are the physical resources backed by GPU memory. GPUCommandBuffer and GPURenderBundle are containers for user-recorded commands. GPUShaderModule contains shader code. The other resources, such as GPUSampler or GPUBindGroup, configure the way physical resources are used by the GPU.

GPUs execute commands encoded in GPUCommandBuffers by feeding data through a pipeline, which is a mix of fixed-function and programmable stages. Programmable stages execute shaders, which are special programs designed to run on GPU hardware. Most of the state of a pipeline is defined by a GPURenderPipeline or a GPUComputePipeline object. The state not included in these pipeline objects is set during encoding with commands, such as beginRenderPass() or setBlendColor().

2. Security considerations

2.1. CPU-based undefined behavior

In order to disallow insecure usage, the range of allowed WebGPU behaviors is defined for any input. An implementation has to validate all the input from the user and only reach the driver with the valid workloads. This document specifies all the error conditions and handling semantics. For example, specifying the same buffer with intersecting ranges in both "source" and "destination" of copyBufferToBuffer() results in GPUCommandEncoder generating an error, and no other operation occurring.

See § 19 Errors & Debugging for more information about error handling.

2.2. GPU-based undefined behavior

2.3. Out-of-bounds access in shaders

In order to prevent the shaders from accessing GPU memory an application doesn’t own, the WebGPU implementation may enable a special mode (called "robust buffer access") in the driver that guarantees that the access is limited to buffer bounds. Alternatively, an implementation may transform the shader code by inserting manual bounds checks.

If the shader attempts to load data outside of physical resource bounds, the implementation is allowed to:

1. return a value at a different location within the resource bounds

2. return a value vector of "(0, 0, 0, X)" with any "X"

3. partially discard the draw or dispatch call

If the shader attempts to write data outside of physical resource bounds, the implementation is allowed to:

1. write the value to a different location within the resource bounds

2. discard the write operation

3. partially discard the draw or dispatch call

2.4. Invalid data

2.5. Driver bugs

2.6. Timing attacks

2.7. Denial of service

2.8. Fingerprinting

3. Convention

3.1. Coordinate Systems

• Y-axis is up in normalized device coordinate (NDC): point(-1.0, -1.0) in NDC is located at the bottom-left corner of NDC. In addition, x and y in NDC should be between -1.0 and 1.0 inclusive, while z in NDC should be between 0.0 and 1.0 inclusive. Vertices out of this range in NDC will not introduce any errors, but they will be clipped.

• Y-axis is down in framebuffer coordinate, viewport coordinate and fragment/pixel coordinate: origin(0, 0) is located at the top-left corner in these coordinate systems.

• Window/present coordinate matches framebuffer coordinate.

• UV of origin(0, 0) in texture coordinate represents the first texel (the lowest byte) in texture memory.

4. Type Definitions

typedef unsigned long long GPUBufferSize;

4.1. Colors and Vectors

dictionary GPUColorDict {
    required double r;
    required double g;
    required double b;
    required double a;
};
typedef (sequence<double> or GPUColorDict) GPUColor;

Note: double is large enough to precisely hold 32-bit signed/unsigned integers and single-precision floats.

dictionary GPUOrigin2DDict {
    unsigned long x = 0;
    unsigned long y = 0;
};
typedef (sequence<unsigned long> or GPUOrigin2DDict) GPUOrigin2D;

dictionary GPUOrigin3DDict {
    unsigned long x = 0;
    unsigned long y = 0;
    unsigned long z = 0;
};
typedef (sequence<unsigned long> or GPUOrigin3DDict) GPUOrigin3D;

dictionary GPUExtent3DDict {
    required unsigned long width;
    required unsigned long height;
    required unsigned long depth;
};
typedef (sequence<unsigned long> or GPUExtent3DDict) GPUExtent3D;

typedef sequence<any> GPUMappedBuffer;

GPUMappedBuffer is always a sequence of 2 elements, of types GPUBuffer and ArrayBuffer, respectively.

4.2. Internal Objects

An internal object is a non-exposed conceptual WebGPU object. Internal objects track the state of an API object and hold any underlying implementation. If the state of a particular internal object can change in parallel from multiple agents, those changes are always atomic with respect to all agents.

Note: An "agent" refers to a JavaScript "thread" (i.e. main thread, or Web Worker).

4.3. WebGPU Interfaces

A WebGPU interface is an exposed interface which encapsulates an internal object. It provides the interface through which the internal object's state is changed.

Any interface which includes GPUObjectBase is a WebGPU interface.

interface mixin GPUObjectBase {
    attribute DOMString? label;
};

label, of type DOMString, nullable

[[device]], of type device, readonly

An internal slot holding the device on which the object exists.

4.4. Object Descriptors

dictionary GPUObjectDescriptorBase {
    DOMString? label;
};

5. Initialization

5.1. Examples

navigator.gpu.requestAdapter().then(adapter => {
    adapter.requestDevice().then(device => {
        // Use 'device' as needed.
    });
}).catch(error => {
    // WebGPU is unsupported, or no adapters or devices are available.
});

5.2. navigator.gpu

[Exposed=Window]
partial interface Navigator {
    [SameObject] readonly attribute GPU gpu;
};

[Exposed=DedicatedWorker]
partial interface WorkerNavigator {
    [SameObject] readonly attribute GPU gpu;
};

The gpu attribute is used to access a GPU object from the main thread or a dedicated worker.

5.3. GPU

[Exposed=Window]
interface GPU {
    // May reject with DOMException  // TODO: DOMException("OperationError")?
    Promise<GPUAdapter> requestAdapter(optional GPURequestAdapterOptions options = {});
};

GPU defines the interface of navigator.gpu, the entry point to WebGPU. It exposes requestAdapter(), for acquiring adapters.

5.4. Adapters

An adapter represents an implementation of WebGPU on the system. Each adapter identifies both an instance of a hardware accelerator (e.g. GPU or CPU) and an instance of a browser’s implementation of WebGPU on top of that accelerator.

If an adapter becomes unavailable, it becomes invalid. Once invalid, it never becomes valid again.

Note: An adapter may be a physical display adapter (GPU), but it could also be a software renderer. A returned adapter could refer to different physical adapters, or to different browser codepaths or system drivers on the same physical adapters. Applications can hold onto multiple adapters at once (via GPUAdapter) (even if some are invalid), and two of these could refer to different instances of the same physical configuration (e.g. if the GPU reset, or were disconnected and reconnected).

5.4.1. GPUAdapter

A GPUAdapter refers to an adapter, and exposes its capabilities (extensions and limits).

interface GPUAdapter {
    readonly attribute DOMString name;
    readonly attribute object extensions;
    //readonly attribute GPULimits limits; Don’t expose higher limits for now.

    // May reject with DOMException  // TODO: DOMException("OperationError")?
    Promise<GPUDevice> requestDevice(optional GPUDeviceDescriptor descriptor = {});
};

GPUAdapter has the following attributes:

name, of type DOMString, readonly

A human-readable name identifying the adapter. The contents are implementation-defined.

extensions, of type object, readonly

A GPUExtensions object which enumerates the extensions supported by the user agent, and whether each extension is supported by the underlying implementation.

• If an extension is not supported by the user agent, it will not be present in the object.

• If an extension is supported by the user agent, but not by the adapter, it will be false.

• If an extension is supported by the user agent and by the adapter, it will be true.

GPUAdapter also has the following internal slots:

[[adapter]], of type adapter, readonly

An internal slot holding the adapter to which this GPUAdapter refers.

5.4.2. Getting an Adapter

To get a GPUAdapter, use requestAdapter().

dictionary GPURequestAdapterOptions {
    GPUPowerPreference powerPreference;
};

GPURequestAdapterOptions provides requirements and hints to the user agent indicating what configuration is suitable for the application.

powerPreference, of type GPUPowerPreference

Provides a hint indicating what class of adapter should be selected from the system’s available adapters.

If unspecified, the user agent decides which class of adapter is most suitable.

Note: This hint may influence which GPU is used in a system with multiple GPUs. For example, a dual-GPU system might have one GPU that consumes less power at the expense of performance.

Note: Depending on the exact hardware configuration, such as attached displays or removable GPUs, the user agent may select different adapters given the same power preference. Typically, given the same hardware configuration and state (e.g. battery percentage) and powerPreference, the user agent should select the same adapter.

enum GPUPowerPreference {
    "low-power",
    "high-performance"
};

"low-power"

Indicates a request to prioritize power savings over performance.

Note: Generally, content should use this if it is unlikely to be constrained by drawing performance; for example, if it renders only one frame per second, draws only relatively simple geometry with simple shaders, or uses a small HTML canvas element. Developers are encouraged to use this value if their content allows, since it may significantly improve battery life on portable devices.

"high-performance"

Indicates a request to prioritize performance over power consumption.

Note: By choosing this value, developers should be aware that, for devices created on the resulting adapter, user agents are more likely to force device loss, in order to save power. Developers are encouraged to only specify this value if they believe it is absolutely necessary, since it may significantly decrease battery life on portable devices.

5.5. Devices

A device is the logical instantiation of an adapter, through which internal objects are created. It can be shared across multiple agents (e.g. dedicated workers).

A device is the exclusive owner of all internal objects created from it: when the device is lost, all objects created from it become invalid.

A device has the following internal slots:

[[adapter]], of type adapter, readonly

The adapter from which this device was created.

When a device is created, its capabilities are set to those requested by the user in requestDevice() (which must be no greater than the capabilities specified by the adapter). These capabilities are enforced upon the usage of objects on the device, even if the underlying adapter can support higher capabilities.

5.5.1. GPUDevice

A GPUDevice refers to a device and exposes the capabilities with which the device was created. It is the top-level object through which WebGPU interfaces are created.

[Exposed=(Window, Worker), Serializable]
interface GPUDevice : EventTarget {
    readonly attribute GPUAdapter adapter;
    readonly attribute object extensions;
    readonly attribute object limits;

    [SameObject] readonly attribute GPUQueue defaultQueue;

    GPUBuffer createBuffer(GPUBufferDescriptor descriptor);
    GPUMappedBuffer createBufferMapped(GPUBufferDescriptor descriptor);
    Promise<GPUMappedBuffer> createBufferMappedAsync(GPUBufferDescriptor descriptor);
    GPUTexture createTexture(GPUTextureDescriptor descriptor);
    GPUSampler createSampler(optional GPUSamplerDescriptor descriptor = {});

    GPUBindGroupLayout createBindGroupLayout(GPUBindGroupLayoutDescriptor descriptor);
    GPUPipelineLayout createPipelineLayout(GPUPipelineLayoutDescriptor descriptor);
    GPUBindGroup createBindGroup(GPUBindGroupDescriptor descriptor);

    GPUShaderModule createShaderModule(GPUShaderModuleDescriptor descriptor);
    GPUComputePipeline createComputePipeline(GPUComputePipelineDescriptor descriptor);
    GPURenderPipeline createRenderPipeline(GPURenderPipelineDescriptor descriptor);

    GPUCommandEncoder createCommandEncoder(optional GPUCommandEncoderDescriptor descriptor = {});
    GPURenderBundleEncoder createRenderBundleEncoder(GPURenderBundleEncoderDescriptor descriptor);
};
GPUDevice includes GPUObjectBase;

GPUDevice has the following attributes:

adapter, of type GPUAdapter, readonly

The GPUAdapter from which this device was created.

extensions, of type object, readonly

A GPUExtensions object exposing the extensions with which this device was created.

limits, of type object, readonly

A GPULimits object exposing the limits with which this device was created.

GPUDevice also has the following internal slots:

[[device]], of type device, readonly

The device that this GPUDevice refers to.

GPUDevice objects are serializable objects.

5.5.2. Getting a Device

To get a GPUDevice, use requestDevice().

dictionary GPUDeviceDescriptor : GPUObjectDescriptorBase {
    GPUExtensions extensions = {};
    GPULimits limits = {};

    // TODO: are other things configurable like queues?
};

extensions, of type GPUExtensions, defaulting to None

The extensions to request of device creation.

limits, of type GPULimits, defaulting to None

The limits to request of device creation.

dictionary GPUExtensions {
    boolean anisotropicFiltering = false;
};

dictionary GPULimits {
    unsigned long maxBindGroups = 4;
    unsigned long maxDynamicUniformBuffersPerPipelineLayout = 8;
    unsigned long maxDynamicStorageBuffersPerPipelineLayout = 4;
    unsigned long maxSampledTexturesPerShaderStage = 16;
    unsigned long maxSamplersPerShaderStage = 16;
    unsigned long maxStorageBuffersPerShaderStage = 4;
    unsigned long maxStorageTexturesPerShaderStage = 4;
    unsigned long maxUniformBuffersPerShaderStage = 12;
};

6. GPUBuffer

A GPUBuffer represents a block of memory that can be used in GPU operations. Data is stored in linear layout, meaning that each byte of the allocation can be addressed by its offset from the start of the GPUBuffer, subject to alignment restrictions depending on the operation. Some GPUBuffers can be mapped which makes the block of memory accessible via an ArrayBuffer called its mapping.

GPUBuffers can be created via the following functions:

• GPUDevice.createBuffer(descriptor) that returns a new unmapped buffer.

• GPUDevice.createBufferMapped(descriptor) that returns a mapped buffer and its mapping.

• GPUDevice.createBufferMappedAsync(descriptor) that returns a Promise of a mapped buffer and its mapping.

[Serializable]
interface GPUBuffer {
    Promise<ArrayBuffer> mapReadAsync();
    Promise<ArrayBuffer> mapWriteAsync();
    void unmap();

    void destroy();
};
GPUBuffer includes GPUObjectBase;

GPUBuffer has the following internal slots:

[[size]] of type GPUBufferSize.

The length of the GPUBuffer allocation in bytes.

[[usage]] of type GPUBufferUsageFlags.

The allowed usages for this GPUBuffer.

[[state]] of type buffer state.

The current state of the GPUBuffer.

Each GPUBuffer has a current buffer state which is one of the following:

• "mapped" where the GPUBuffer is available for CPU operations.

• "unmapped" where the GPUBuffer is available for GPU operations.

• "destroyed" where the GPUBuffer is no longer available for any operations except destroy.

Note: [[size]] and [[usage]] are immutable once the GPUBuffer has been created.

GPUBuffer is Serializable. It is a reference to an internal buffer object, and Serializable means that the reference can be copied between realms (threads/workers), allowing multiple realms to access it concurrently. Since GPUBuffer has internal state (mapped, destroyed), that state is internally-synchronized - these state changes occur atomically across realms.

6.1. Buffer creation

6.1.1. GPUBufferDescriptor

This specifies the options to use in creating a GPUBuffer.

dictionary GPUBufferDescriptor : GPUObjectDescriptorBase {
    required GPUBufferSize size;
    required GPUBufferUsageFlags usage;
};

validating GPUBufferDescriptor(device, descriptor)

1. If device is lost return false.

2. If any of the bits of descriptor’s usage aren’t present in this device’s [[allowed buffer usages]] return false.

3. If both the MAP_READ and MAP_WRITE bits of descriptor’s usage attribute are set, return false.

4. Return true.

6.1.2. GPUDevice.createBuffer(descriptor)

createBuffer(descriptor)

1. If the result of validating GPUBufferDescriptor(this, descriptor) is false:

   1. Record a validation error in the current scope.

   2. Create an invalid GPUBuffer and return the result.

2. Let b be a new GPUBuffer object.

3. Set the [[size]] slot of b to the value of the size attribute of descriptor.

4. Set the [[usage]] slot of b to the value of the usage attribute of descriptor.

5. Set the [[state]] internal slot of b to unmapped.

6. Set each byte of b’s allocation to zero.

7. Return b.

6.2. Buffer Destruction

An application that no longer requires a GPUBuffer can choose to lose access to it before garbage collection by calling destroy().

Note: This allows the user agent to reclaim the GPU memory associated with the GPUBuffer once all previously submitted operations using it are complete.

destroy()

1. If the [[state]] slot of this is mapped

   1. Run the steps to unmap "this"

2. Set the [[state]] slot of this to destroyed

6.3. Buffer Usage

typedef unsigned long GPUBufferUsageFlags;
interface GPUBufferUsage {
    const GPUBufferUsageFlags MAP_READ  = 0x0001;
    const GPUBufferUsageFlags MAP_WRITE = 0x0002;
    const GPUBufferUsageFlags COPY_SRC  = 0x0004;
    const GPUBufferUsageFlags COPY_DST  = 0x0008;
    const GPUBufferUsageFlags INDEX     = 0x0010;
    const GPUBufferUsageFlags VERTEX    = 0x0020;
    const GPUBufferUsageFlags UNIFORM   = 0x0040;
    const GPUBufferUsageFlags STORAGE   = 0x0080;
    const GPUBufferUsageFlags INDIRECT  = 0x0100;
};

6.4. Buffer Mapping

7. Textures

7.1. GPUTexture

[Serializable]
interface GPUTexture {
    GPUTextureView createView(optional GPUTextureViewDescriptor descriptor = {});

    void destroy();
};
GPUTexture includes GPUObjectBase;

7.1.1. Texture Creation

dictionary GPUTextureDescriptor : GPUObjectDescriptorBase {
    required GPUExtent3D size;
    unsigned long arrayLayerCount = 1;
    unsigned long mipLevelCount = 1;
    unsigned long sampleCount = 1;
    GPUTextureDimension dimension = "2d";
    required GPUTextureFormat format;
    required GPUTextureUsageFlags usage;
};

enum GPUTextureDimension {
    "1d",
    "2d",
    "3d"
};

typedef unsigned long GPUTextureUsageFlags;
interface GPUTextureUsage {
    const GPUTextureUsageFlags COPY_SRC          = 0x01;
    const GPUTextureUsageFlags COPY_DST          = 0x02;
    const GPUTextureUsageFlags SAMPLED           = 0x04;
    const GPUTextureUsageFlags STORAGE           = 0x08;
    const GPUTextureUsageFlags OUTPUT_ATTACHMENT = 0x10;
};

7.2. GPUTextureView

interface GPUTextureView {
};
GPUTextureView includes GPUObjectBase;

7.2.1. Texture View Creation

dictionary GPUTextureViewDescriptor : GPUObjectDescriptorBase {
    GPUTextureFormat format;
    GPUTextureViewDimension dimension;
    GPUTextureAspect aspect = "all";
    unsigned long baseMipLevel = 0;
    unsigned long mipLevelCount = 0;
    unsigned long baseArrayLayer = 0;
    unsigned long arrayLayerCount = 0;
};

enum GPUTextureViewDimension {
    "1d",
    "2d",
    "2d-array",
    "cube",
    "cube-array",
    "3d"
};

enum GPUTextureAspect {
    "all",
    "stencil-only",
    "depth-only"
};

7.3. Texture Formats

The name of the format specifies the order of components, bits per component, and data type for the component.

• r, g, b, a = red, green, blue, alpha

• unorm = unsigned normalized

• snorm = signed normalized

• uint = unsigned int

• sint = signed int

• float = floating point

If the format has the -srgb suffix, then sRGB gamma compression and decompression are applied during the reading and writing of color values in the pixel. Compressed texture formats are provided by extensions. Their naming should follow the convention here, with the texture name as a prefix. e.g. etc2-rgba8unorm.

enum GPUTextureFormat {
    // 8-bit formats
    "r8unorm",
    "r8snorm",
    "r8uint",
    "r8sint",

    // 16-bit formats
    "r16uint",
    "r16sint",
    "r16float",
    "rg8unorm",
    "rg8snorm",
    "rg8uint",
    "rg8sint",

    // 32-bit formats
    "r32uint",
    "r32sint",
    "r32float",
    "rg16uint",
    "rg16sint",
    "rg16float",
    "rgba8unorm",
    "rgba8unorm-srgb",
    "rgba8snorm",
    "rgba8uint",
    "rgba8sint",
    "bgra8unorm",
    "bgra8unorm-srgb",
    // Packed 32-bit formats
    "rgb10a2unorm",
    "rg11b10float",

    // 64-bit formats
    "rg32uint",
    "rg32sint",
    "rg32float",
    "rgba16uint",
    "rgba16sint",
    "rgba16float",

    // 128-bit formats
    "rgba32uint",
    "rgba32sint",
    "rgba32float",

    // Depth and stencil formats
    "depth32float",
    "depth24plus",
    "depth24plus-stencil8"
};

• The depth24plus family of formats (depth24plus and depth24plus-stencil8) must have a depth-component precision of 1 ULP ≤ 1 / (2²⁴).

  Note: This is unlike the 24-bit unsigned normalized format family typically found in native APIs, which has a precision of 1 ULP = 1 / (2²⁴ − 1).

enum GPUTextureComponentType {
    "float",
    "sint",
    "uint"
};

8. Samplers

8.1. GPUSampler

interface GPUSampler {
};
GPUSampler includes GPUObjectBase;

8.1.1. Creation

dictionary GPUSamplerDescriptor : GPUObjectDescriptorBase {
    GPUAddressMode addressModeU = "clamp-to-edge";
    GPUAddressMode addressModeV = "clamp-to-edge";
    GPUAddressMode addressModeW = "clamp-to-edge";
    GPUFilterMode magFilter = "nearest";
    GPUFilterMode minFilter = "nearest";
    GPUFilterMode mipmapFilter = "nearest";
    float lodMinClamp = 0;
    float lodMaxClamp = 0xffffffff; // TODO: What should this be? Was Number.MAX_VALUE.
    GPUCompareFunction compare = "never";
};

enum GPUAddressMode {
    "clamp-to-edge",
    "repeat",
    "mirror-repeat"
};

enum GPUFilterMode {
    "nearest",
    "linear"
};

enum GPUCompareFunction {
    "never",
    "less",
    "equal",
    "less-equal",
    "greater",
    "not-equal",
    "greater-equal",
    "always"
};

9. Resource Binding

9.1. GPUBindGroupLayout

A GPUBindGroupLayout defines the interface between a set of resources bound in a GPUBindGroup and their accessibility in shader stages.

[Serializable]
interface GPUBindGroupLayout {
};
GPUBindGroupLayout includes GPUObjectBase;

9.1.1. Creation

A GPUBindGroupLayout is created via GPUDevice.createBindGroupLayout().

dictionary GPUBindGroupLayoutDescriptor : GPUObjectDescriptorBase {
    required sequence<GPUBindGroupLayoutBinding> bindings;
};

A GPUBindGroupLayoutBinding describes a single shader resource binding to be included in a GPUBindGroupLayout.

dictionary GPUBindGroupLayoutBinding {
    required unsigned long binding;
    required GPUShaderStageFlags visibility;
    required GPUBindingType type;
    GPUTextureViewDimension textureDimension = "2d";
    GPUTextureComponentType textureComponentType = "float";
    boolean multisampled = false;
    boolean hasDynamicOffset = false;
};

• binding: A unique identifier for a resource binding within a GPUBindGroupLayoutBinding, a corresponding GPUBindGroupBinding, and shader stages.

• visibility: A bitset of the members of GPUShaderStage. Each set bit indicates that a GPUBindGroupLayoutBinding's resource will be accessible from the associated shader stage.

typedef unsigned long GPUShaderStageFlags;
interface GPUShaderStage {
    const GPUShaderStageFlags VERTEX   = 0x1;
    const GPUShaderStageFlags FRAGMENT = 0x2;
    const GPUShaderStageFlags COMPUTE  = 0x4;
};

• type: A member of GPUBindingType that indicates the intended usage of a resource binding in its visible GPUShaderStages.

enum GPUBindingType {
    "uniform-buffer",
    "storage-buffer",
    "readonly-storage-buffer",
    "sampler",
    "sampled-texture",
    "storage-texture"
    // TODO: other binding types
};

• textureDimension, multisampled: Describes the dimensionality of texture view bindings, and indicates if they are multisampled.

  Note: This allows Metal-based implementations to back the respective bind groups with MTLArgumentBuffer objects that are more efficient to bind at run-time.

• hasDynamicOffset: For uniform-buffer, storage-buffer, and readonly-storage-buffer bindings, indicates that the binding has a dynamic offset. One offset must be passed to setBindGroup for each dynamic binding in increasing order of binding number.

A GPUBindGroupLayout object has the following internal slots:

[[bindings]] of type sequence<GPUBindGroupLayoutBinding>.

The set of GPUBindGroupLayoutBindings this GPUBindGroupLayout describes.

9.1.2. GPUDevice.createBindGroupLayout(GPUBindGroupLayoutDescriptor)

9.2. GPUBindGroup

interface GPUBindGroup {
};
GPUBindGroup includes GPUObjectBase;

9.2.1. Bind Group Creation

dictionary GPUBindGroupDescriptor : GPUObjectDescriptorBase {
    required GPUBindGroupLayout layout;
    required sequence<GPUBindGroupBinding> bindings;
};

typedef (GPUSampler or GPUTextureView or GPUBufferBinding) GPUBindingResource;

dictionary GPUBindGroupBinding {
    required unsigned long binding;
    required GPUBindingResource resource;
};

dictionary GPUBufferBinding {
    required GPUBuffer buffer;
    GPUBufferSize offset = 0;
    GPUBufferSize size;
};

• size: If undefined, specifies the range starting at offset and ending at the end of the buffer.

9.3. GPUPipelineLayout

interface GPUPipelineLayout {
};
GPUPipelineLayout includes GPUObjectBase;

9.3.1. Creation

dictionary GPUPipelineLayoutDescriptor : GPUObjectDescriptorBase {
    required sequence<GPUBindGroupLayout> bindGroupLayouts;
};

10. Shader Modules

10.1. GPUShaderModule

[Serializable]
interface GPUShaderModule {
};
GPUShaderModule includes GPUObjectBase;

GPUShaderModule is Serializable. It is a reference to an internal shader module object, and Serializable means that the reference can be copied between realms (threads/workers), allowing multiple realms to access it concurrently. Since GPUShaderModule immutable, there are no race conditions.

10.1.1. Shader Module Creation

typedef (Uint32Array or DOMString) GPUShaderCode;

dictionary GPUShaderModuleDescriptor : GPUObjectDescriptorBase {
    required GPUShaderCode code;
};

Note: While the choice of shader language is undecided, GPUShaderModuleDescriptor will temporarily accept both text and binary input.

11. Pipelines

dictionary GPUPipelineDescriptorBase : GPUObjectDescriptorBase {
    required GPUPipelineLayout layout;
};

dictionary GPUProgrammableStageDescriptor {
    required GPUShaderModule module;
    required DOMString entryPoint;
    // TODO: other stuff like specialization constants?
};

11.1. GPUComputePipeline

[Serializable]
interface GPUComputePipeline {
};
GPUComputePipeline includes GPUObjectBase;

11.1.1. Creation

dictionary GPUComputePipelineDescriptor : GPUPipelineDescriptorBase {
    required GPUProgrammableStageDescriptor computeStage;
};

11.2. GPURenderPipeline

[Serializable]
interface GPURenderPipeline {
};
GPURenderPipeline includes GPUObjectBase;

11.2.1. Creation

dictionary GPURenderPipelineDescriptor : GPUPipelineDescriptorBase {
    required GPUProgrammableStageDescriptor vertexStage;
    GPUProgrammableStageDescriptor fragmentStage;

    required GPUPrimitiveTopology primitiveTopology;
    GPURasterizationStateDescriptor rasterizationState = {};
    required sequence<GPUColorStateDescriptor> colorStates;
    GPUDepthStencilStateDescriptor depthStencilState;
    GPUVertexStateDescriptor vertexState = {};

    unsigned long sampleCount = 1;
    unsigned long sampleMask = 0xFFFFFFFF;
    boolean alphaToCoverageEnabled = false;
    // TODO: other properties
};

• sampleCount: Number of MSAA samples.

11.2.2. Primitive Topology

enum GPUPrimitiveTopology {
    "point-list",
    "line-list",
    "line-strip",
    "triangle-list",
    "triangle-strip"
};

11.2.3. Rasterization State

dictionary GPURasterizationStateDescriptor {
    GPUFrontFace frontFace = "ccw";
    GPUCullMode cullMode = "none";

    long depthBias = 0;
    float depthBiasSlopeScale = 0;
    float depthBiasClamp = 0;
};

enum GPUFrontFace {
    "ccw",
    "cw"
};

enum GPUCullMode {
    "none",
    "front",
    "back"
};

11.2.4. Color State

dictionary GPUColorStateDescriptor {
    required GPUTextureFormat format;

    GPUBlendDescriptor alphaBlend = {};
    GPUBlendDescriptor colorBlend = {};
    GPUColorWriteFlags writeMask = 0xF;  // GPUColorWrite.ALL
};

typedef unsigned long GPUColorWriteFlags;
interface GPUColorWrite {
    const GPUColorWriteFlags RED   = 0x1;
    const GPUColorWriteFlags GREEN = 0x2;
    const GPUColorWriteFlags BLUE  = 0x4;
    const GPUColorWriteFlags ALPHA = 0x8;
    const GPUColorWriteFlags ALL   = 0xF;
};

11.2.4.1. Blend State

dictionary GPUBlendDescriptor {
    GPUBlendFactor srcFactor = "one";
    GPUBlendFactor dstFactor = "zero";
    GPUBlendOperation operation = "add";
};

enum GPUBlendFactor {
    "zero",
    "one",
    "src-color",
    "one-minus-src-color",
    "src-alpha",
    "one-minus-src-alpha",
    "dst-color",
    "one-minus-dst-color",
    "dst-alpha",
    "one-minus-dst-alpha",
    "src-alpha-saturated",
    "blend-color",
    "one-minus-blend-color"
};

enum GPUBlendOperation {
    "add",
    "subtract",
    "reverse-subtract",
    "min",
    "max"
};

enum GPUStencilOperation {
    "keep",
    "zero",
    "replace",
    "invert",
    "increment-clamp",
    "decrement-clamp",
    "increment-wrap",
    "decrement-wrap"
};

11.2.5. Depth/Stencil State

dictionary GPUDepthStencilStateDescriptor {
    required GPUTextureFormat format;

    boolean depthWriteEnabled = false;
    GPUCompareFunction depthCompare = "always";

    GPUStencilStateFaceDescriptor stencilFront = {};
    GPUStencilStateFaceDescriptor stencilBack = {};

    unsigned long stencilReadMask = 0xFFFFFFFF;
    unsigned long stencilWriteMask = 0xFFFFFFFF;
};

dictionary GPUStencilStateFaceDescriptor {
    GPUCompareFunction compare = "always";
    GPUStencilOperation failOp = "keep";
    GPUStencilOperation depthFailOp = "keep";
    GPUStencilOperation passOp = "keep";
};

11.2.6. Vertex State

enum GPUIndexFormat {
    "uint16",
    "uint32"
};

11.2.6.1. Vertex formats

The name of the format specifies the data type of the component, the number of values, and whether the data is normalized.

• uchar = unsigned 8-bit value

• char = signed 8-bit value

• ushort = unsigned 16-bit value

• short = signed 16-bit value

• half = half-precision 16-bit floating point value

• float = 32-bit floating point value

• uint = unsigned 32-bit integer value

• int = signed 32-bit integer value

If no number of values is given in the name, a single value is provided. If the format has the -bgra suffix, it means the values are arranged as blue, green, red and alpha values.

enum GPUVertexFormat {
    "uchar2",
    "uchar4",
    "char2",
    "char4",
    "uchar2norm",
    "uchar4norm",
    "char2norm",
    "char4norm",
    "ushort2",
    "ushort4",
    "short2",
    "short4",
    "ushort2norm",
    "ushort4norm",
    "short2norm",
    "short4norm",
    "half2",
    "half4",
    "float",
    "float2",
    "float3",
    "float4",
    "uint",
    "uint2",
    "uint3",
    "uint4",
    "int",
    "int2",
    "int3",
    "int4"
};

enum GPUInputStepMode {
    "vertex",
    "instance"
};

dictionary GPUVertexStateDescriptor {
    GPUIndexFormat indexFormat = "uint32";
    sequence<GPUVertexBufferLayoutDescriptor?> vertexBuffers = [];
};

A vertex buffer is, conceptually, a view into buffer memory as an array of structures. arrayStride is the stride, in bytes, between elements of that array. Each element of a vertex buffer is like a structure with a memory layout defined by its attributes, which describe the members of the structure.

Each GPUVertexAttributeDescriptor describes its format and its offset, in bytes, within the structure.

Each attribute appears as a separate input in a vertex shader, each bound by a numeric location, which is specified by shaderLocation. Every location must be unique within the GPUVertexStateDescriptor.

dictionary GPUVertexBufferLayoutDescriptor {
    required GPUBufferSize arrayStride;
    GPUInputStepMode stepMode = "vertex";
    required sequence<GPUVertexAttributeDescriptor> attributes;
};

dictionary GPUVertexAttributeDescriptor {
    required GPUVertexFormat format;
    required GPUBufferSize offset;

    required unsigned long shaderLocation;
};

12. Command Buffers

12.1. GPUCommandBuffer

interface GPUCommandBuffer {
};
GPUCommandBuffer includes GPUObjectBase;

12.1.1. Creation

dictionary GPUCommandBufferDescriptor : GPUObjectDescriptorBase {
};

13. Command Encoding

13.1. GPUCommandEncoder

interface GPUCommandEncoder {
    GPURenderPassEncoder beginRenderPass(GPURenderPassDescriptor descriptor);
    GPUComputePassEncoder beginComputePass(optional GPUComputePassDescriptor descriptor = {});

    void copyBufferToBuffer(
        GPUBuffer source,
        GPUBufferSize sourceOffset,
        GPUBuffer destination,
        GPUBufferSize destinationOffset,
        GPUBufferSize size);

    void copyBufferToTexture(
        GPUBufferCopyView source,
        GPUTextureCopyView destination,
        GPUExtent3D copySize);

    void copyTextureToBuffer(
        GPUTextureCopyView source,
        GPUBufferCopyView destination,
        GPUExtent3D copySize);

    void copyTextureToTexture(
        GPUTextureCopyView source,
        GPUTextureCopyView destination,
        GPUExtent3D copySize);

    void copyImageBitmapToTexture(
        GPUImageBitmapCopyView source,
        GPUTextureCopyView destination,
        GPUExtent3D copySize);

    void pushDebugGroup(DOMString groupLabel);
    void popDebugGroup();
    void insertDebugMarker(DOMString markerLabel);

    GPUCommandBuffer finish(optional GPUCommandBufferDescriptor descriptor = {});
};
GPUCommandEncoder includes GPUObjectBase;

• copyImageBitmapToTexture():

  • For now, copySize.z must be 1.

13.1.1. Creation

dictionary GPUCommandEncoderDescriptor : GPUObjectDescriptorBase {
    // TODO: reusability flag?
};

13.2. Copy Commands

dictionary GPUBufferCopyView {
    required GPUBuffer buffer;
    GPUBufferSize offset = 0;
    required unsigned long rowPitch;
    required unsigned long imageHeight;
};

dictionary GPUTextureCopyView {
    required GPUTexture texture;
    unsigned long mipLevel = 0;
    unsigned long arrayLayer = 0;
    GPUOrigin3D origin = {};
};

• origin: If unspecified, defaults to [0, 0, 0].

dictionary GPUImageBitmapCopyView {
    required ImageBitmap imageBitmap;
    GPUOrigin2D origin = {};
};

• origin: If unspecified, defaults to [0, 0].

13.3. Programmable Passes

interface mixin GPUProgrammablePassEncoder {
    void setBindGroup(unsigned long index, GPUBindGroup bindGroup,
                      optional sequence<unsigned long> dynamicOffsets = []);

    void setBindGroup(unsigned long index, GPUBindGroup bindGroup,
                      Uint32Array dynamicOffsetsData,
                      unsigned long long dynamicOffsetsDataStart,
                      unsigned long long dynamicOffsetsDataLength);

    void pushDebugGroup(DOMString groupLabel);
    void popDebugGroup();
    void insertDebugMarker(DOMString markerLabel);
};

Debug groups in a GPUCommandEncoder or GPUProgrammablePassEncoder must be well nested.

14. Compute Passes

14.1. GPUComputePassEncoder

interface GPUComputePassEncoder {
    void setPipeline(GPUComputePipeline pipeline);
    void dispatch(unsigned long x, optional unsigned long y = 1, optional unsigned long z = 1);
    void dispatchIndirect(GPUBuffer indirectBuffer, GPUBufferSize indirectOffset);

    void endPass();
};
GPUComputePassEncoder includes GPUObjectBase;
GPUComputePassEncoder includes GPUProgrammablePassEncoder;

14.1.1. Creation

dictionary GPUComputePassDescriptor : GPUObjectDescriptorBase {
};

15. Render Passes

15.1. GPURenderPassEncoder

interface mixin GPURenderEncoderBase {
    void setPipeline(GPURenderPipeline pipeline);

    void setIndexBuffer(GPUBuffer buffer, optional GPUBufferSize offset = 0);
    void setVertexBuffer(unsigned long slot, GPUBuffer buffer, optional GPUBufferSize offset = 0);

    void draw(unsigned long vertexCount, unsigned long instanceCount,
              unsigned long firstVertex, unsigned long firstInstance);
    void drawIndexed(unsigned long indexCount, unsigned long instanceCount,
                     unsigned long firstIndex, long baseVertex, unsigned long firstInstance);

    void drawIndirect(GPUBuffer indirectBuffer, GPUBufferSize indirectOffset);
    void drawIndexedIndirect(GPUBuffer indirectBuffer, GPUBufferSize indirectOffset);
};

interface GPURenderPassEncoder {
    void setViewport(float x, float y,
                     float width, float height,
                     float minDepth, float maxDepth);

    void setScissorRect(unsigned long x, unsigned long y, unsigned long width, unsigned long height);

    void setBlendColor(GPUColor color);
    void setStencilReference(unsigned long reference);

    void executeBundles(sequence<GPURenderBundle> bundles);
    void endPass();
};
GPURenderPassEncoder includes GPUObjectBase;
GPURenderPassEncoder includes GPUProgrammablePassEncoder;
GPURenderPassEncoder includes GPURenderEncoderBase;

• In indirect draw calls, the base instance field (inside the indirect buffer data) must be set to zero.

• setScissorRect():

  • An error is generated if width or height is not greater than 0.

When a GPURenderPassEncoder is created, it has the following default state:

• Viewport:

  • x, y = 0.0, 0.0

  • width, height = the dimensions of the pass’s render targets

  • minDepth, maxDepth = 0.0, 1.0

• Scissor rectangle:

  • x, y = 0, 0

  • width, height = the dimensions of the pass’s render targets

When a GPURenderBundle is executed, it does not inherit the pass’s pipeline, bind groups, or vertex or index buffers. After a GPURenderBundle has executed, the pass’s pipeline, bind groups, and vertex and index buffers are cleared. If zero GPURenderBundles are executed, the command buffer state is unchanged.

15.1.1. Creation

dictionary GPURenderPassDescriptor : GPUObjectDescriptorBase {
    required sequence<GPURenderPassColorAttachmentDescriptor> colorAttachments;
    GPURenderPassDepthStencilAttachmentDescriptor depthStencilAttachment;
};

15.1.1.1. Color Attachments

dictionary GPURenderPassColorAttachmentDescriptor {
    required GPUTextureView attachment;
    GPUTextureView resolveTarget;

    required (GPULoadOp or GPUColor) loadValue;
    GPUStoreOp storeOp = "store";
};

15.1.1.2. Depth/Stencil Attachments

dictionary GPURenderPassDepthStencilAttachmentDescriptor {
    required GPUTextureView attachment;

    required (GPULoadOp or float) depthLoadValue;
    required GPUStoreOp depthStoreOp;

    required (GPULoadOp or unsigned long) stencilLoadValue;
    required GPUStoreOp stencilStoreOp;
};

15.1.2. Load & Store Operations

enum GPULoadOp {
    "load"
};

enum GPUStoreOp {
    "store",
    "clear"
};

16. Bundles

16.1. GPURenderBundle

interface GPURenderBundle {
};
GPURenderBundle includes GPUObjectBase;

16.1.1. Creation

dictionary GPURenderBundleDescriptor : GPUObjectDescriptorBase {
};

interface GPURenderBundleEncoder {
    GPURenderBundle finish(optional GPURenderBundleDescriptor descriptor = {});
};
GPURenderBundleEncoder includes GPUObjectBase;
GPURenderBundleEncoder includes GPUProgrammablePassEncoder;
GPURenderBundleEncoder includes GPURenderEncoderBase;

16.1.2. Encoding

dictionary GPURenderBundleEncoderDescriptor : GPUObjectDescriptorBase {
    required sequence<GPUTextureFormat> colorFormats;
    GPUTextureFormat depthStencilFormat;
    unsigned long sampleCount = 1;
};

17. Queues

interface GPUQueue {
    void submit(sequence<GPUCommandBuffer> commandBuffers);

    GPUFence createFence(optional GPUFenceDescriptor descriptor = {});
    void signal(GPUFence fence, unsigned long long signalValue);
};
GPUQueue includes GPUObjectBase;

submit(commandBuffers) does nothing and produces an error if any of the following is true:

• Any GPUBuffer referenced in any element of commandBuffers isn’t in the "unmapped" buffer state.

17.1. GPUFence

interface GPUFence {
    unsigned long long getCompletedValue();
    Promise<void> onCompletion(unsigned long long completionValue);
};
GPUFence includes GPUObjectBase;

17.1.1. Creation

dictionary GPUFenceDescriptor : GPUObjectDescriptorBase {
    unsigned long long initialValue = 0;
};

18. Canvas Rendering and Swap Chain

interface GPUCanvasContext {
    GPUSwapChain configureSwapChain(GPUSwapChainDescriptor descriptor);

    Promise<GPUTextureFormat> getSwapChainPreferredFormat(GPUDevice device);
};

• configureSwapChain(): Configures the swap chain for this canvas, and returns a new GPUSwapChain object representing it. Destroys any swapchain previously returned by configureSwapChain, including all of the textures it has produced.

dictionary GPUSwapChainDescriptor : GPUObjectDescriptorBase {
    required GPUDevice device;
    required GPUTextureFormat format;
    GPUTextureUsageFlags usage = 0x10;  // GPUTextureUsage.OUTPUT_ATTACHMENT
};

interface GPUSwapChain {
    GPUTexture getCurrentTexture();
};
GPUSwapChain includes GPUObjectBase;

19. Errors & Debugging

19.1. Fatal Errors

interface GPUDeviceLostInfo {
    readonly attribute DOMString message;
};

partial interface GPUDevice {
    readonly attribute Promise<GPUDeviceLostInfo> lost;
};

19.2. Error Scopes

enum GPUErrorFilter {
    "none",
    "out-of-memory",
    "validation"
};

interface GPUOutOfMemoryError {
    constructor();
};

interface GPUValidationError {
    constructor(DOMString message);
    readonly attribute DOMString message;
};

typedef (GPUOutOfMemoryError or GPUValidationError) GPUError;

partial interface GPUDevice {
    void pushErrorScope(GPUErrorFilter filter);
    Promise<GPUError?> popErrorScope();
};

popErrorScope() throws OperationError if there are no error scopes on the stack. popErrorScope() rejects with OperationError if the device is lost.

19.3. Telemetry

[
    Exposed=Window
]
interface GPUUncapturedErrorEvent : Event {
    constructor(
        DOMString type,
        GPUUncapturedErrorEventInit gpuUncapturedErrorEventInitDict
    );
    readonly attribute GPUError error;
};

dictionary GPUUncapturedErrorEventInit : EventInit {
    required GPUError error;
};

partial interface GPUDevice {
    [Exposed=Window]
    attribute EventHandler onuncapturederror;
};

20. Temporary usages of non-exported dfns #

Eventually all of these should disappear but they are useful to avoid warning while building the specification.

vertex buffer