Struct wgpu_types::Features

#[repr(transparent)]
pub struct Features { /* private fields */ }

Features that are not guaranteed to be supported.

These are either part of the webgpu standard, or are extension features supported by wgpu when targeting native.

If you want to use a feature, you need to first verify that the adapter supports the feature. If the adapter does not support the feature, requesting a device with it enabled will panic.

Corresponds to WebGPU GPUFeatureName.

Implementations

impl Features

pub const DEPTH_CLIP_CONTROL: Self = _

By default, polygon depth is clipped to 0-1 range before/during rasterization. Anything outside of that range is rejected, and respective fragments are not touched.

With this extension, we can disabling clipping. That allows shadow map occluders to be rendered into a tighter depth range.

Supported platforms:

• desktops
• some mobile chips

This is a web and native feature.

pub const DEPTH32FLOAT_STENCIL8: Self = _

Allows for explicit creation of textures of format TextureFormat::Depth32FloatStencil8

Supported platforms:

• Vulkan (mostly)
• DX12
• Metal

This is a web and native feature.

pub const TEXTURE_COMPRESSION_BC: Self = _

Enables BCn family of compressed textures. All BCn textures use 4x4 pixel blocks with 8 or 16 bytes per block.

Compressed textures sacrifice some quality in exchange for significantly reduced bandwidth usage.

Support for this feature guarantees availability of [TextureUsages::COPY_SRC | TextureUsages::COPY_DST | TextureUsages::TEXTURE_BINDING] for BCn formats. Features::TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES may enable additional usages.

Supported Platforms:

• desktops

This is a web and native feature.

pub const TEXTURE_COMPRESSION_ETC2: Self = _

Enables ETC family of compressed textures. All ETC textures use 4x4 pixel blocks. ETC2 RGB and RGBA1 are 8 bytes per block. RTC2 RGBA8 and EAC are 16 bytes per block.

Compressed textures sacrifice some quality in exchange for significantly reduced bandwidth usage.

Support for this feature guarantees availability of [TextureUsages::COPY_SRC | TextureUsages::COPY_DST | TextureUsages::TEXTURE_BINDING] for ETC2 formats. Features::TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES may enable additional usages.

Supported Platforms:

• Vulkan on Intel
• Mobile (some)

This is a web and native feature.

pub const TEXTURE_COMPRESSION_ASTC_LDR: Self = _

Enables ASTC family of compressed textures. ASTC textures use pixel blocks varying from 4x4 to 12x12. Blocks are always 16 bytes.

Compressed textures sacrifice some quality in exchange for significantly reduced bandwidth usage.

Support for this feature guarantees availability of [TextureUsages::COPY_SRC | TextureUsages::COPY_DST | TextureUsages::TEXTURE_BINDING] for ASTC formats. Features::TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES may enable additional usages.

Supported Platforms:

• Vulkan on Intel
• Mobile (some)

This is a web and native feature.

pub const INDIRECT_FIRST_INSTANCE: Self = _

Allows non-zero value for the “first instance” in indirect draw calls.

Supported Platforms:

• Vulkan (mostly)
• DX12
• Metal

This is a web and native feature.

pub const TIMESTAMP_QUERY: Self = _

Enables use of Timestamp Queries. These queries tell the current gpu timestamp when all work before the query is finished. Call [CommandEncoder::write_timestamp], [RenderPassEncoder::write_timestamp], or [ComputePassEncoder::write_timestamp] to write out a timestamp.

They must be resolved using [CommandEncoder::resolve_query_sets] into a buffer, then the result must be multiplied by the timestamp period [Queue::get_timestamp_period] to get the timestamp in nanoseconds. Multiple timestamps can then be diffed to get the time for operations between them to finish.

Supported Platforms:

• Vulkan
• DX12

This is a web and native feature.

pub const PIPELINE_STATISTICS_QUERY: Self = _

Enables use of Pipeline Statistics Queries. These queries tell the count of various operations performed between the start and stop call. Call [RenderPassEncoder::begin_pipeline_statistics_query] to start a query, then call [RenderPassEncoder::end_pipeline_statistics_query] to stop one.

They must be resolved using [CommandEncoder::resolve_query_sets] into a buffer. The rules on how these resolve into buffers are detailed in the documentation for PipelineStatisticsTypes.

Supported Platforms:

• Vulkan
• DX12

This is a web and native feature.

pub const SHADER_FLOAT16: Self = _

Allows shaders to acquire the FP16 ability

Note: this is not supported in naga yet，only through spir-v passthrough right now.

Supported Platforms:

• Vulkan
• Metal

This is a web and native feature.

pub const MAPPABLE_PRIMARY_BUFFERS: Self = _

Webgpu only allows the MAP_READ and MAP_WRITE buffer usage to be matched with COPY_DST and COPY_SRC respectively. This removes this requirement.

This is only beneficial on systems that share memory between CPU and GPU. If enabled on a system that doesn’t, this can severely hinder performance. Only use if you understand the consequences.

Supported platforms:

• Vulkan
• DX12
• Metal

This is a native only feature.

pub const TEXTURE_BINDING_ARRAY: Self = _

Allows the user to create uniform arrays of textures in shaders:

ex.

• var textures: binding_array<texture_2d<f32>, 10> (WGSL)
• uniform texture2D textures[10] (GLSL)

If Features::STORAGE_RESOURCE_BINDING_ARRAY is supported as well as this, the user may also create uniform arrays of storage textures.

ex.

• var textures: array<texture_storage_2d<f32, write>, 10> (WGSL)
• uniform image2D textures[10] (GLSL)

This capability allows them to exist and to be indexed by dynamically uniform values.

Supported platforms:

• DX12
• Metal (with MSL 2.0+ on macOS 10.13+)
• Vulkan

This is a native only feature.

pub const BUFFER_BINDING_ARRAY: Self = _

Allows the user to create arrays of buffers in shaders:

ex.

• var<uniform> buffer_array: array<MyBuffer, 10> (WGSL)
• uniform myBuffer { ... } buffer_array[10] (GLSL)

This capability allows them to exist and to be indexed by dynamically uniform values.

If Features::STORAGE_RESOURCE_BINDING_ARRAY is supported as well as this, the user may also create arrays of storage buffers.

ex.

• var<storage> buffer_array: array<MyBuffer, 10> (WGSL)
• buffer myBuffer { ... } buffer_array[10] (GLSL)

Supported platforms:

• DX12
• Vulkan

This is a native only feature.

pub const STORAGE_RESOURCE_BINDING_ARRAY: Self = _

Allows the user to create uniform arrays of storage buffers or textures in shaders, if resp. Features::BUFFER_BINDING_ARRAY or Features::TEXTURE_BINDING_ARRAY is supported.

This capability allows them to exist and to be indexed by dynamically uniform values.

Supported platforms:

• Metal (with MSL 2.2+ on macOS 10.13+)
• Vulkan

This is a native only feature.

pub const SAMPLED_TEXTURE_AND_STORAGE_BUFFER_ARRAY_NON_UNIFORM_INDEXING: Self = _

Allows shaders to index sampled texture and storage buffer resource arrays with dynamically non-uniform values:

ex. texture_array[vertex_data]

In order to use this capability, the corresponding GLSL extension must be enabled like so:

#extension GL_EXT_nonuniform_qualifier : require

and then used either as nonuniformEXT qualifier in variable declaration:

ex. layout(location = 0) nonuniformEXT flat in int vertex_data;

or as nonuniformEXT constructor:

ex. texture_array[nonuniformEXT(vertex_data)]

WGSL and HLSL do not need any extension.

Supported platforms:

• DX12
• Metal (with MSL 2.0+ on macOS 10.13+)
• Vulkan 1.2+ (or VK_EXT_descriptor_indexing)’s shaderSampledImageArrayNonUniformIndexing & shaderStorageBufferArrayNonUniformIndexing feature)

This is a native only feature.

pub const UNIFORM_BUFFER_AND_STORAGE_TEXTURE_ARRAY_NON_UNIFORM_INDEXING: Self = _

Allows shaders to index uniform buffer and storage texture resource arrays with dynamically non-uniform values:

ex. texture_array[vertex_data]

In order to use this capability, the corresponding GLSL extension must be enabled like so:

#extension GL_EXT_nonuniform_qualifier : require

and then used either as nonuniformEXT qualifier in variable declaration:

ex. layout(location = 0) nonuniformEXT flat in int vertex_data;

or as nonuniformEXT constructor:

ex. texture_array[nonuniformEXT(vertex_data)]

WGSL and HLSL do not need any extension.

Supported platforms:

• DX12
• Metal (with MSL 2.0+ on macOS 10.13+)
• Vulkan 1.2+ (or VK_EXT_descriptor_indexing)’s shaderUniformBufferArrayNonUniformIndexing & shaderStorageTextureArrayNonUniformIndexing feature)

This is a native only feature.

pub const PARTIALLY_BOUND_BINDING_ARRAY: Self = _

Allows the user to create bind groups continaing arrays with less bindings than the BindGroupLayout.

This is a native only feature.

pub const MULTI_DRAW_INDIRECT: Self = _

Allows the user to call [RenderPass::multi_draw_indirect] and [RenderPass::multi_draw_indexed_indirect].

Allows multiple indirect calls to be dispatched from a single buffer.

Supported platforms:

• DX12
• Vulkan
• Metal (Emulated on top of draw_indirect and draw_indexed_indirect)

This is a native only feature.

pub const MULTI_DRAW_INDIRECT_COUNT: Self = _

Allows the user to call [RenderPass::multi_draw_indirect_count] and [RenderPass::multi_draw_indexed_indirect_count].

This allows the use of a buffer containing the actual number of draw calls.

Supported platforms:

• DX12
• Vulkan 1.2+ (or VK_KHR_draw_indirect_count)

This is a native only feature.

pub const PUSH_CONSTANTS: Self = _

Allows the use of push constants: small, fast bits of memory that can be updated inside a [RenderPass].

Allows the user to call [RenderPass::set_push_constants], provide a non-empty array to [PipelineLayoutDescriptor], and provide a non-zero limit to Limits::max_push_constant_size.

A block of push constants can be declared with layout(push_constant) uniform Name {..} in shaders.

Supported platforms:

• DX12
• Vulkan
• Metal
• DX11 (emulated with uniforms)
• OpenGL (emulated with uniforms)

This is a native only feature.

pub const ADDRESS_MODE_CLAMP_TO_BORDER: Self = _

Allows the use of AddressMode::ClampToBorder with a border color other than SamplerBorderColor::Zero.

Supported platforms:

• DX12
• Vulkan
• Metal (macOS 10.12+ only)
• DX11
• OpenGL

This is a web and native feature.

pub const POLYGON_MODE_LINE: Self = _

Allows the user to set PolygonMode::Line in PrimitiveState::polygon_mode

This allows drawing polygons/triangles as lines (wireframe) instead of filled

Supported platforms:

• DX12
• Vulkan
• Metal

This is a native only feature.

pub const POLYGON_MODE_POINT: Self = _

Allows the user to set PolygonMode::Point in PrimitiveState::polygon_mode

This allows only drawing the vertices of polygons/triangles instead of filled

Supported platforms:

• DX12
• Vulkan

This is a native only feature.

pub const TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES: Self = _

Enables device specific texture format features.

See TextureFormatFeatures for a listing of the features in question.

By default only texture format properties as defined by the WebGPU specification are allowed. Enabling this feature flag extends the features of each format to the ones supported by the current device. Note that without this flag, read/write storage access is not allowed at all.

This extension does not enable additional formats.

This is a native-only feature.

pub const SHADER_FLOAT64: Self = _

Enables 64-bit floating point types in SPIR-V shaders.

Note: even when supported by GPU hardware, 64-bit floating point operations are frequently between 16 and 64 times slower than equivalent operations on 32-bit floats.

Supported Platforms:

• Vulkan

This is a native-only feature.

pub const VERTEX_ATTRIBUTE_64BIT: Self = _

Enables using 64-bit types for vertex attributes.

Requires SHADER_FLOAT64.

Supported Platforms: N/A

This is a native-only feature.

pub const CONSERVATIVE_RASTERIZATION: Self = _

Allows the user to set a overestimation-conservative-rasterization in PrimitiveState::conservative

Processing of degenerate triangles/lines is hardware specific. Only triangles are supported.

Supported platforms:

• Vulkan

This is a native only feature.

pub const VERTEX_WRITABLE_STORAGE: Self = _

Enables bindings of writable storage buffers and textures visible to vertex shaders.

Note: some (tiled-based) platforms do not support vertex shaders with any side-effects.

Supported Platforms:

• All

This is a native-only feature.

pub const CLEAR_TEXTURE: Self = _

Enables clear to zero for textures.

Supported platforms:

• All

This is a native only feature.

pub const SPIRV_SHADER_PASSTHROUGH: Self = _

Enables creating shader modules from SPIR-V binary data (unsafe).

SPIR-V data is not parsed or interpreted in any way; you can use [wgpu::make_spirv_raw!] to check for alignment and magic number when converting from raw bytes.

Supported platforms:

• Vulkan, in case shader’s requested capabilities and extensions agree with Vulkan implementation.

This is a native only feature.

pub const SHADER_PRIMITIVE_INDEX: Self = _

Enables builtin(primitive_index) in fragment shaders.

Note: enables geometry processing for pipelines using the builtin. This may come with a significant performance impact on some hardware. Other pipelines are not affected.

Supported platforms:

• Vulkan
• DX11 (feature level 10+)
• DX12
• Metal (some)
• OpenGL (some)

This is a native only feature.

pub const MULTIVIEW: Self = _

Enables multiview render passes and builtin(view_index) in vertex shaders.

Supported platforms:

• Vulkan
• OpenGL (web only)

This is a native only feature.

pub const TEXTURE_FORMAT_16BIT_NORM: Self = _

Enables normalized 16-bit texture formats.

Supported platforms:

• Vulkan
• DX12
• Metal

This is a native only feature.

pub const ADDRESS_MODE_CLAMP_TO_ZERO: Self = _

Allows the use of AddressMode::ClampToBorder with a border color of SamplerBorderColor::Zero.

Supported platforms:

• DX12
• Vulkan
• Metal
• DX11
• OpenGL

This is a native only feature.

pub const TEXTURE_COMPRESSION_ASTC_HDR: Self = _

Enables ASTC HDR family of compressed textures.

Compressed textures sacrifice some quality in exchange for significantly reduced bandwidth usage.

Support for this feature guarantees availability of [TextureUsages::COPY_SRC | TextureUsages::COPY_DST | TextureUsages::TEXTURE_BINDING] for BCn formats. Features::TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES may enable additional usages.

Supported Platforms:

• Metal
• Vulkan
• OpenGL

This is a native-only feature.

pub const WRITE_TIMESTAMP_INSIDE_PASSES: Self = _

Allows for timestamp queries inside render passes. Metal does not allow this on Apple GPUs.

Implies [Features::TIMESTAMP_QUERIES] is supported.

Supported platforms:

• Vulkan
• DX12
• Metal (Intel and AMD GPUs)

pub const fn empty() -> Self

Returns an empty set of flags.

pub const fn all() -> Self

Returns the set containing all flags.

pub const fn bits(&self) -> u64

Returns the raw value of the flags currently stored.

pub const fn from_bits(bits: u64) -> Option<Self>

Convert from underlying bit representation, unless that representation contains bits that do not correspond to a flag.

pub const fn from_bits_truncate(bits: u64) -> Self

Convert from underlying bit representation, dropping any bits that do not correspond to flags.

pub const unsafe fn from_bits_unchecked(bits: u64) -> Self

Convert from underlying bit representation, preserving all bits (even those not corresponding to a defined flag).

Safety

The caller of the bitflags! macro can chose to allow or disallow extra bits for their bitflags type.

The caller of from_bits_unchecked() has to ensure that all bits correspond to a defined flag or that extra bits are valid for this bitflags type.

pub const fn is_empty(&self) -> bool

Returns true if no flags are currently stored.

pub const fn is_all(&self) -> bool

Returns true if all flags are currently set.

pub const fn intersects(&self, other: Self) -> bool

Returns true if there are flags common to both self and other.

pub const fn contains(&self, other: Self) -> bool

Returns true if all of the flags in other are contained within self.

pub fn insert(&mut self, other: Self)

Inserts the specified flags in-place.

pub fn remove(&mut self, other: Self)

Removes the specified flags in-place.

pub fn toggle(&mut self, other: Self)

Toggles the specified flags in-place.

pub fn set(&mut self, other: Self, value: bool)

Inserts or removes the specified flags depending on the passed value.

pub const fn intersection(self, other: Self) -> Self

Returns the intersection between the flags in self and other.

Specifically, the returned set contains only the flags which are present in both self and other.

This is equivalent to using the & operator (e.g. ops::BitAnd), as in flags & other.

pub const fn union(self, other: Self) -> Self

Returns the union of between the flags in self and other.

Specifically, the returned set contains all flags which are present in either self or other, including any which are present in both (see Self::symmetric_difference if that is undesirable).

This is equivalent to using the | operator (e.g. ops::BitOr), as in flags | other.

pub const fn difference(self, other: Self) -> Self

Returns the difference between the flags in self and other.

Specifically, the returned set contains all flags present in self, except for the ones present in other.

It is also conceptually equivalent to the “bit-clear” operation: flags & !other (and this syntax is also supported).

This is equivalent to using the - operator (e.g. ops::Sub), as in flags - other.

pub const fn symmetric_difference(self, other: Self) -> Self

Returns the symmetric difference between the flags in self and other.

Specifically, the returned set contains the flags present which are present in self or other, but that are not present in both. Equivalently, it contains the flags present in exactly one of the sets self and other.

This is equivalent to using the ^ operator (e.g. ops::BitXor), as in flags ^ other.

pub const fn complement(self) -> Self

Returns the complement of this set of flags.

Specifically, the returned set contains all the flags which are not set in self, but which are allowed for this type.

Alternatively, it can be thought of as the set difference between Self::all() and self (e.g. Self::all() - self)

This is equivalent to using the ! operator (e.g. ops::Not), as in !flags.

impl Features

pub fn contains_invalid_bits(&self) -> bool

Returns true if the bitflags contains bits that are not part of the bitflags definition.

impl Features

pub const fn all_webgpu_mask() -> Self

Mask of all features which are part of the upstream WebGPU standard.

pub const fn all_native_mask() -> Self

Mask of all features that are only available when targeting native (not web).

Trait Implementations

impl Binary for Features

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl BitAnd<Features> for Features

fn bitand(self, other: Self) -> Self

Returns the intersection between the two sets of flags.

type Output = Features

The resulting type after applying the & operator.

impl BitAndAssign<Features> for Features

fn bitand_assign(&mut self, other: Self)

Disables all flags disabled in the set.

impl BitOr<Features> for Features

fn bitor(self, other: Features) -> Self

Returns the union of the two sets of flags.

type Output = Features

The resulting type after applying the | operator.

impl BitOrAssign<Features> for Features

fn bitor_assign(&mut self, other: Self)

Adds the set of flags.

impl BitXor<Features> for Features

fn bitxor(self, other: Self) -> Self

Returns the left flags, but with all the right flags toggled.

type Output = Features

The resulting type after applying the ^ operator.

impl BitXorAssign<Features> for Features

fn bitxor_assign(&mut self, other: Self)

Toggles the set of flags.

impl Clone for Features

fn clone(&self) -> Features

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Features

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for Features

fn default() -> Features

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for Features

Available on crate feature replay only.

fn deserialize<D>(deserializer: D) -> Result<Features, D::Error>where D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Extend<Features> for Features

fn extend<T: IntoIterator<Item = Self>>(&mut self, iterator: T)

Extends a collection with the contents of an iterator.

fn extend_one(&mut self, item: A)

🔬This is a nightly-only experimental API. (extend_one)

Extends a collection with exactly one element.

fn extend_reserve(&mut self, additional: usize)

🔬This is a nightly-only experimental API. (extend_one)

Reserves capacity in a collection for the given number of additional elements.

impl FromIterator<Features> for Features

fn from_iter<T: IntoIterator<Item = Self>>(iterator: T) -> Self

Creates a value from an iterator.

impl Hash for Features

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl LowerHex for Features

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Not for Features

fn not(self) -> Self

Returns the complement of this set of flags.

type Output = Features

The resulting type after applying the ! operator.

impl Octal for Features

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Ord for Features

fn cmp(&self, other: &Features) -> Ordering

This method returns an Ordering between self and other.

fn max(self, other: Self) -> Selfwhere Self: Sized,

Compares and returns the maximum of two values.

fn min(self, other: Self) -> Selfwhere Self: Sized,

Compares and returns the minimum of two values.

fn clamp(self, min: Self, max: Self) -> Selfwhere Self: Sized + PartialOrd<Self>,

Restrict a value to a certain interval.

impl PartialEq<Features> for Features

fn eq(&self, other: &Features) -> bool

This method tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

This method tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl PartialOrd<Features> for Features

fn partial_cmp(&self, other: &Features) -> Option<Ordering>

This method returns an ordering between self and other values if one exists.

fn lt(&self, other: &Rhs) -> bool

This method tests less than (for self and other) and is used by the < operator.

fn le(&self, other: &Rhs) -> bool

This method tests less than or equal to (for self and other) and is used by the <= operator.

fn gt(&self, other: &Rhs) -> bool

This method tests greater than (for self and other) and is used by the > operator.

fn ge(&self, other: &Rhs) -> bool

This method tests greater than or equal to (for self and other) and is used by the >= operator.

impl Serialize for Features

Available on crate feature trace only.

fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>where S: Serializer,

Serialize this value into the given Serde serializer.

impl Sub<Features> for Features

fn sub(self, other: Self) -> Self

Returns the set difference of the two sets of flags.

type Output = Features

The resulting type after applying the - operator.

impl SubAssign<Features> for Features

fn sub_assign(&mut self, other: Self)

Disables all flags enabled in the set.

impl UpperHex for Features

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Copy for Features

impl Eq for Features

impl StructuralEq for Features

impl StructuralPartialEq for Features