Struct TierCaps 

pub struct TierCaps {

    pub downlevel_flags: DownlevelFlags,
    pub adapter_name: String,
    pub backend: Backend,
    pub max_texture_dimension_2d: u32,
    pub max_texture_array_layers: u32,
    pub max_bind_groups: u32,
    pub max_uniform_buffer_binding_size: u32,
    pub min_uniform_buffer_offset_alignment: u32,
    pub max_vertex_attributes: u32,
    pub has_storage_buffers: bool,
    pub has_timestamp_query: bool,
    pub transient_saves_memory: bool,
    pub atlas_format: TextureFormat,
    pub resource_texture_dim: u32,
    pub downlevel_profile: DownlevelProfile,
}

Plain-data adapter capabilities the GPU crate’s device/pipeline/atlas decisions are built over. Constructed from a real wgpu::Adapter via TierCaps::probe, or by hand in host tests via TierCaps::fake.

Fields

downlevel_flags: DownlevelFlags

The adapter’s wgpu::DownlevelCapabilities::flags.

adapter_name: String

The adapter’s wgpu::AdapterInfo::name, used only for diagnostics.

backend: Backend

The adapter’s wgpu::AdapterInfo::backend.

max_texture_dimension_2d: u32

The adapter’s wgpu::Limits::max_texture_dimension_2d, clamped down to wgpu::Limits::downlevel_webgl2_defaults().max_texture_dimension_2d when Self::downlevel_profile is DownlevelProfile::WebGl2.

max_texture_array_layers: u32

The adapter’s wgpu::Limits::max_texture_array_layers, clamped the same way as Self::max_texture_dimension_2d.

max_bind_groups: u32

The adapter’s wgpu::Limits::max_bind_groups, clamped the same way as Self::max_texture_dimension_2d.

max_uniform_buffer_binding_size: u32

The adapter’s wgpu::Limits::max_uniform_buffer_binding_size, saturated into a u32 (the field is u64 on wgpu::Limits; no adapter this workspace targets reports a value anywhere near u32::MAX) and clamped the same way as Self::max_texture_dimension_2d.

min_uniform_buffer_offset_alignment: u32

The adapter’s wgpu::Limits::min_uniform_buffer_offset_alignment, raised to wgpu::Limits::downlevel_webgl2_defaults().min_uniform_buffer_offset_alignment (256) when Self::downlevel_profile is DownlevelProfile::WebGl2 and the adapter reports a looser (smaller) value — an alignment requirement is a floor, so it is only ever raised, never lowered.

max_vertex_attributes: u32

The adapter’s wgpu::Limits::max_vertex_attributes, clamped the same way as Self::max_texture_dimension_2d.

has_storage_buffers: bool

Whether the adapter supports storage buffers at all (wgpu::Limits::max_storage_buffers_per_shader_stage > 0) — forced false whenever Self::downlevel_profile is DownlevelProfile::WebGl2, since the GLES-3.0/WebGL2 downlevel default limits report zero storage buffers per shader stage.

has_timestamp_query: bool

Whether the adapter exposes wgpu::Features::TIMESTAMP_QUERY.

transient_saves_memory: bool

The adapter’s wgpu::AdapterInfo::transient_saves_memory — whether adding wgpu::TextureUsages::TRANSIENT_ATTACHMENT to a texture (which itself requires wgpu::StoreOp::Discard) reduces memory usage on this adapter.

wgpu reports this as Option<bool>, where None means “the adapter does not say” (only the web backend, which no frust shell targets). Self::probe folds None to false: an unknown answer must not buy the transient flag, since it is a memory optimization with a hard clear/discard precondition and no upside when the driver ignores it.

atlas_format: TextureFormat

The texture format a texture atlas is backed by.

resource_texture_dim: u32

The texture dimension a pooled/atlas resource texture is sized against: min(max_texture_dimension_2d, 4096) against the (possibly already WebGL2-clamped) Self::max_texture_dimension_2d, so a very generous desktop adapter’s real ceiling never drives an oversized allocation.

downlevel_profile: DownlevelProfile

Whether this adapter’s usable limits are the full native set or clamped to the GLES-3.0/WebGL2 downlevel defaults.

Implementations

impl TierCaps

pub fn probe(adapter: &Adapter) -> Self

Probes a real wgpu::Adapter for the capabilities this crate’s device/pipeline/atlas decisions are built over. Synchronous: every value it reads (get_info, get_downlevel_capabilities, limits, features) is available before device creation.

When the resolved DownlevelProfile is DownlevelProfile::WebGl2 — a real wgpu::Backend::Gl adapter, or FRUST_ENGINE_DOWNLEVEL=1 rehearsing it against a desktop backend — the probed limits are run through clamp_to_webgl2_defaults and storage buffers are forced off, so the override actually rehearses the downlevel shape instead of only relabelling desktop values.

pub fn fake(profile: DownlevelProfile) -> Self

Builds synthetic caps for the given DownlevelProfile with no wgpu::Adapter at all — the host-test counterpart to Self::probe.

Full mirrors wgpu::Limits::defaults()’s desktop values (backend Vulkan, max_texture_dimension_2d 8192, storage buffers present); WebGl2 mirrors wgpu::Limits::downlevel_webgl2_defaults() (backend Gl, max_texture_dimension_2d 2048, no storage buffers).

Trait Implementations

impl Clone for TierCaps

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TierCaps

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

Formats the value using the given formatter.

impl Eq for TierCaps

impl PartialEq for TierCaps

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

Equality operator ==.

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

Inequality operator !=.

impl StructuralPartialEq for TierCaps