pub struct Instance { /* private fields */ }
Expand description
Context for all other wgpu objects. Instance of wgpu.
This is the first thing you create when using wgpu.
Its primary use is to create Adapter
s and Surface
s.
Does not have to be kept alive.
Corresponds to WebGPU GPU
.
Implementations§
source§impl Instance
impl Instance
sourcepub const fn enabled_backend_features() -> Backends
pub const fn enabled_backend_features() -> Backends
Returns which backends can be picked for the current build configuration.
The returned set depends on a combination of target platform and enabled features. This does not do any runtime checks and is exclusively based on compile time information.
InstanceDescriptor::backends
does not need to be a subset of this,
but any backend that is not in this set, will not be picked.
TODO: Right now it’s otherwise not possible yet to opt-out of all features on some platforms. See https://github.com/gfx-rs/wgpu/issues/3514
- Windows/Linux/Android: always enables Vulkan and GLES with no way to opt out
sourcepub fn new(_instance_desc: InstanceDescriptor) -> Self
pub fn new(_instance_desc: InstanceDescriptor) -> Self
Create an new instance of wgpu.
§Arguments
-
instance_desc
- Has fields for which backends wgpu will choose during instantiation, and which DX12 shader compiler wgpu will use.Backends::BROWSER_WEBGPU
takes a special role: If it is set and WebGPU support is detected, this instance will only be able to create WebGPU adapters. If you instead want to force use of WebGL, either disable thewebgpu
compile-time feature or do add theBackends::BROWSER_WEBGPU
flag to the theinstance_desc
’sbackends
field. If it is set and WebGPU support is not detected, the instance will use wgpu-core to create adapters. Meaning that if thewebgl
feature is enabled, it is able to create a WebGL adapter.
§Panics
If no backend feature for the active target platform is enabled,
this method will panic, see Instance::enabled_backend_features()
.
sourcepub fn enumerate_adapters(&self, backends: Backends) -> Vec<Adapter>
pub fn enumerate_adapters(&self, backends: Backends) -> Vec<Adapter>
sourcepub fn request_adapter(
&self,
options: &RequestAdapterOptions<'_, '_>
) -> impl Future<Output = Option<Adapter>> + WasmNotSend
pub fn request_adapter( &self, options: &RequestAdapterOptions<'_, '_> ) -> impl Future<Output = Option<Adapter>> + WasmNotSend
Retrieves an Adapter
which matches the given RequestAdapterOptions
.
Some options are “soft”, so treated as non-mandatory. Others are “hard”.
If no adapters are found that suffice all the “hard” options, None
is returned.
sourcepub unsafe fn create_adapter_from_hal<A: HalApi>(
&self,
hal_adapter: ExposedAdapter<A>
) -> Adapter
pub unsafe fn create_adapter_from_hal<A: HalApi>( &self, hal_adapter: ExposedAdapter<A> ) -> Adapter
sourcepub fn create_surface<'window>(
&self,
target: impl Into<SurfaceTarget<'window>>
) -> Result<Surface<'window>, CreateSurfaceError>
pub fn create_surface<'window>( &self, target: impl Into<SurfaceTarget<'window>> ) -> Result<Surface<'window>, CreateSurfaceError>
Creates a new surface targeting a given window/canvas/surface/etc..
Internally, this creates surfaces for all backends that are enabled for this instance.
See SurfaceTarget
for what targets are supported.
See Instance::create_surface_unsafe
for surface creation with unsafe target variants.
Most commonly used are window handles (or provider of windows handles)
which can be passed directly as they’re automatically converted to SurfaceTarget
.
sourcepub unsafe fn create_surface_unsafe<'window>(
&self,
target: SurfaceTargetUnsafe
) -> Result<Surface<'window>, CreateSurfaceError>
pub unsafe fn create_surface_unsafe<'window>( &self, target: SurfaceTargetUnsafe ) -> Result<Surface<'window>, CreateSurfaceError>
Creates a new surface targeting a given window/canvas/surface/etc. using an unsafe target.
Internally, this creates surfaces for all backends that are enabled for this instance.
See SurfaceTargetUnsafe
for what targets are supported.
See Instance::create_surface
for surface creation with safe target variants.
§Safety
- See respective
SurfaceTargetUnsafe
variants for safety requirements.
sourcepub fn poll_all(&self, force_wait: bool) -> bool
pub fn poll_all(&self, force_wait: bool) -> bool
Polls all devices.
If force_wait
is true and this is not running on the web, then this
function will block until all in-flight buffers have been mapped and
all submitted commands have finished execution.
Return true
if all devices’ queues are empty, or false
if there are
queue submissions still in flight. (Note that, unless access to all
Queue
s associated with this Instance
is coordinated somehow,
this information could be out of date by the time the caller receives
it. Queue
s can be shared between threads, and other threads could
submit new work at any time.)
On the web, this is a no-op. Device
s are automatically polled.
sourcepub fn generate_report(&self) -> Option<GlobalReport>
pub fn generate_report(&self) -> Option<GlobalReport>
Generates memory report.
Returns None
if the feature is not supported by the backend
which happens only when WebGPU is pre-selected by the instance creation.
Trait Implementations§
source§impl Default for Instance
impl Default for Instance
source§fn default() -> Self
fn default() -> Self
Creates a new instance of wgpu with default options.
Backends are set to Backends::all()
, and FXC is chosen as the dx12_shader_compiler
.
§Panics
If no backend feature for the active target platform is enabled,
this method will panic, see Instance::enabled_backend_features()
.