Struct FallbackContext 

pub struct FallbackContext { /* private fields */ }

A GpuContext wrapper that carries a FallbackConfig and provides fallback-aware operation dispatch.

Before invoking the GPU closure, FallbackContext checks the device-lost flag. If the device is already known to be lost, it routes directly to the CPU closure without attempting any GPU work.

Example

use oxigdal_gpu::{GpuContext, GpuError};
use oxigdal_gpu::cpu_fallback::{FallbackConfig, FallbackContext, ExecutionPath};

let gpu = GpuContext::new().await?;
let fb = FallbackContext::with_default_config(gpu);

let result = fb.execute(
    || Ok::<u32, GpuError>(1_u32),
    || 2_u32,
)?;

assert_eq!(result.path, ExecutionPath::Gpu);

Implementations

impl FallbackContext

pub fn new(ctx: GpuContext, config: FallbackConfig) -> Self

Create a new FallbackContext with the supplied configuration.

pub fn with_default_config(ctx: GpuContext) -> Self

Create a FallbackContext with the default FallbackConfig.

pub fn gpu_context(&self) -> &GpuContext

Return a reference to the underlying GpuContext.

pub fn fallback_config(&self) -> &FallbackConfig

Return a reference to the active FallbackConfig.

pub fn is_device_lost(&self) -> bool

Return true if the GPU device has been lost.

pub fn execute<T, GpuFn, CpuFn>( &self, gpu_op: GpuFn, cpu_op: CpuFn, ) -> GpuResult<FallbackResult<T>>

where GpuFn: FnOnce() -> GpuResult<T>, CpuFn: FnOnce() -> T,

Execute gpu_op with automatic CPU fallback per the configured policy.

If the device-lost flag is already set, the GPU closure is skipped entirely and cpu_op is called directly (provided fallback is enabled). Otherwise the behaviour mirrors execute_with_fallback.

Trait Implementations

impl Debug for FallbackContext

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

Formats the value using the given formatter.