Struct WgpuContext 

pub struct WgpuContext {
    pub instance: Instance,
    pub adapter: Adapter,
    pub device: Arc<Device>,
    pub queue: Arc<Queue>,
    pub pool: Arc<BufferPool>,
}

A GPU device, its queue and a buffer pool shared by tensors.

Fields

instance: Instance

The wgpu instance the adapter came from.

adapter: Adapter

The physical GPU.

device: Arc<Device>

Logical device; requested with the adapter’s full limits.

queue: Arc<Queue>

Queue that every kernel dispatch is submitted to.

pool: Arc<BufferPool>

Pool that UnifiedTensor::empty draws from.

Implementations

impl WgpuContext

pub async fn new() -> Self

Opens the high-performance adapter.

Panics

Panics if no adapter or device is available.

pub async fn new_with_options( power_preference: PowerPreference, force_fallback_adapter: bool, ) -> Self

Opens an adapter with the given power preference, optionally forcing the software fallback adapter.

Panics

Panics if no adapter or device is available.

pub fn adapter_info(&self) -> String

Adapter name and backend, e.g. "Apple M2 (Metal)".

Trait Implementations

impl Debug for WgpuContext

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

Formats the value using the given formatter.