pub struct AsyncFullStridedLoading {}Expand description
Loads the content of all the stage using all planes, keeping the original layout, making each tile strided
Trait Implementations§
Source§impl Clone for AsyncFullStridedLoading
impl Clone for AsyncFullStridedLoading
Source§fn clone(&self) -> AsyncFullStridedLoading
fn clone(&self) -> AsyncFullStridedLoading
Returns a duplicate of the value. Read more
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
Performs copy-assignment from
source. Read moreSource§impl CubeType for AsyncFullStridedLoading
impl CubeType for AsyncFullStridedLoading
Source§impl FullLoadingStrategy<RuntimeArgs> for AsyncFullStridedLoading
impl FullLoadingStrategy<RuntimeArgs> for AsyncFullStridedLoading
Source§type TilingLayout = StridedTilingLayout
type TilingLayout = StridedTilingLayout
The layout describing how data is tiled across the stage.
Source§type SyncStrategy = AsyncCopy
type SyncStrategy = AsyncCopy
The synchronization strategy that should be used with this loading strategy
Source§type Job<EG: Numeric, NG: Size, ES: Numeric, NS: Size> = AsyncFullStridedJob
type Job<EG: Numeric, NG: Size, ES: Numeric, NS: Size> = AsyncFullStridedJob
The LoadingJob for this strategy.
type Stage = StridedStageFamily
type TileKind = Strided
Source§fn new_job<EG: Numeric, NG: Size, ES: Numeric, NS: Size>(
runtime_args: RuntimeArgs,
config: GlobalReaderConfig,
) -> Self::Job<EG, NG, ES, NS>
fn new_job<EG: Numeric, NG: Size, ES: Numeric, NS: Size>( runtime_args: RuntimeArgs, config: GlobalReaderConfig, ) -> Self::Job<EG, NG, ES, NS>
Returns the job with preliminary calculations done.
fn __expand_new_job<EG: Numeric, NG: Size, ES: Numeric, NS: Size>( scope: &mut Scope, runtime_args: <RuntimeArgs as CubeType>::ExpandType, config: GlobalReaderConfig, ) -> <Self::Job<EG, NG, ES, NS> as CubeType>::ExpandType
const SHOULD_CLEAR: bool = false
Source§impl LoadMaxRoundPlaneCount for AsyncFullStridedLoading
impl LoadMaxRoundPlaneCount for AsyncFullStridedLoading
Source§fn max_round_plane_count(
elements_per_tile: u32,
tiles_per_stage: u32,
vector_size: VectorSize,
plane_dim: u32,
dtype: StorageType,
) -> u32
fn max_round_plane_count( elements_per_tile: u32, tiles_per_stage: u32, vector_size: VectorSize, plane_dim: u32, dtype: StorageType, ) -> u32
Returns the largest number of planes that evenly divides the tasks.
Source§impl LoadingValidation for AsyncFullStridedLoading
impl LoadingValidation for AsyncFullStridedLoading
Source§fn validate_with_config(
device_props: &DeviceProperties,
config: &GlobalReaderConfig,
) -> Result<(), InvalidConfigError>
fn validate_with_config( device_props: &DeviceProperties, config: &GlobalReaderConfig, ) -> Result<(), InvalidConfigError>
Verify that configs are valid for a reader, otherwise return an error stating why
fn validate_with_problem( problem: &MatmulProblem, dtypes: &MatmulElems, ident: StageIdent, ) -> Result<(), InvalidConfigError>
impl Copy for AsyncFullStridedLoading
Auto Trait Implementations§
impl Freeze for AsyncFullStridedLoading
impl RefUnwindSafe for AsyncFullStridedLoading
impl Send for AsyncFullStridedLoading
impl Sync for AsyncFullStridedLoading
impl Unpin for AsyncFullStridedLoading
impl UnsafeUnpin for AsyncFullStridedLoading
impl UnwindSafe for AsyncFullStridedLoading
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Mutably borrows from an owned value. Read more