Struct StridedTilingLayout

pub struct StridedTilingLayout {}

Tiles follow a strided layout that often mirrors global memory layout. Not all tiles are contiguous in shared memory, but mapping is more direct.

Implementations

impl StridedTilingLayout

pub fn nth_slice<ES: Numeric, S: StageConfig>( stage: &mut StageMemory<ES, Self>, nth: u32, ident: Ident, config: S, ) -> SliceMut<Line<ES>>

Returns the nth slice of the stage

pub fn __expand_nth_slice<ES: Numeric, S: StageConfig>( scope: &mut Scope, stage: <StageMemory<ES, Self> as CubeType>::ExpandType, nth: <u32 as CubeType>::ExpandType, ident: Ident, config: S, ) -> <SliceMut<Line<ES>> as CubeType>::ExpandType

Trait Implementations

impl<MP: MatmulPrecision> AsyncLoadingJob<MP, StridedTilingLayout> for AsynFullMaximizeSliceLengthJob

fn execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( this: &mut Self, task_id: u32, tensor_reader: &TensorReader<MP::EI>, stage: &mut StageMemory<MP::ES, StridedTilingLayout>, mechanism: &CM, config: G, )

Execute the task_idth loading task

fn task_count(this: &Self) -> u32

Get the number of tasks

fn __expand_execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( scope: &mut Scope, this: <Self as CubeType>::ExpandType, task_id: <u32 as CubeType>::ExpandType, tensor_reader: <TensorReader<MP::EI> as CubeType>::ExpandType, stage: <StageMemory<MP::ES, StridedTilingLayout> as CubeType>::ExpandType, mechanism: <CM as CubeType>::ExpandType, config: G, ) -> <() as CubeType>::ExpandType

fn __expand_task_count( scope: &mut Scope, this: <Self as CubeType>::ExpandType, ) -> u32

impl<MP: MatmulPrecision> AsyncLoadingJob<MP, StridedTilingLayout> for AsyncFullCooperativeJob

fn execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( this: &mut Self, task_id: u32, tensor_reader: &TensorReader<MP::EI>, stage: &mut StageMemory<MP::ES, StridedTilingLayout>, mechanism: &CM, config: G, )

Execute the task_idth loading task

fn task_count(this: &Self) -> u32

Get the number of tasks

fn __expand_execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( scope: &mut Scope, this: <Self as CubeType>::ExpandType, task_id: <u32 as CubeType>::ExpandType, tensor_reader: <TensorReader<MP::EI> as CubeType>::ExpandType, stage: <StageMemory<MP::ES, StridedTilingLayout> as CubeType>::ExpandType, mechanism: <CM as CubeType>::ExpandType, config: G, ) -> <() as CubeType>::ExpandType

fn __expand_task_count( scope: &mut Scope, this: <Self as CubeType>::ExpandType, ) -> u32

impl<MP: MatmulPrecision> AsyncLoadingJob<MP, StridedTilingLayout> for AsyncFullMaximizeUnitCountJob

fn execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( this: &mut Self, _task_id: u32, tensor_reader: &TensorReader<MP::EI>, stage: &mut StageMemory<MP::ES, StridedTilingLayout>, mechanism: &CM, config: G, )

Execute the task_idth loading task

fn task_count(_this: &Self) -> u32

Get the number of tasks

fn __expand_execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( scope: &mut Scope, this: <Self as CubeType>::ExpandType, _task_id: <u32 as CubeType>::ExpandType, tensor_reader: <TensorReader<MP::EI> as CubeType>::ExpandType, stage: <StageMemory<MP::ES, StridedTilingLayout> as CubeType>::ExpandType, mechanism: <CM as CubeType>::ExpandType, config: G, ) -> <() as CubeType>::ExpandType

fn __expand_task_count( scope: &mut Scope, _this: <Self as CubeType>::ExpandType, ) -> u32

impl<MP: MatmulPrecision> AsyncLoadingJob<MP, StridedTilingLayout> for AsyncPartialMaximizeSliceLengthJob

fn execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( this: &mut Self, task_id: u32, tensor_reader: &TensorReader<MP::EI>, stage: &mut StageMemory<MP::ES, StridedTilingLayout>, mechanism: &CM, config: G, )

Execute the task_idth loading task

fn task_count(this: &Self) -> u32

Get the number of tasks

fn __expand_execute_task<CM: CopyMechanism<MP::ES>, G: GlobalConfig>( scope: &mut Scope, this: <Self as CubeType>::ExpandType, task_id: <u32 as CubeType>::ExpandType, tensor_reader: <TensorReader<MP::EI> as CubeType>::ExpandType, stage: <StageMemory<MP::ES, StridedTilingLayout> as CubeType>::ExpandType, mechanism: <CM as CubeType>::ExpandType, config: G, ) -> <() as CubeType>::ExpandType

fn __expand_task_count( scope: &mut Scope, this: <Self as CubeType>::ExpandType, ) -> u32

impl Clone for StridedTilingLayout

fn clone(&self) -> StridedTilingLayout

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<MP: MatmulPrecision> LoadingJob<MP, StridedTilingLayout> for SyncFullStridedJob

fn execute_task<G: GlobalConfig>( this: &mut Self, task_id: u32, tensor_reader: &TensorReader<MP::EI>, stage: &mut StageMemory<MP::ES, StridedTilingLayout>, quantization: &CubeOption<Quantization<MP>>, config: G, )

Execute the task_idth loading task

fn task_count(this: &Self) -> u32

Get the number of tasks

fn __expand_execute_task<G: GlobalConfig>( scope: &mut Scope, this: <Self as CubeType>::ExpandType, task_id: u32, tensor_reader: <TensorReader<MP::EI> as CubeType>::ExpandType, stage: <StageMemory<MP::ES, StridedTilingLayout> as CubeType>::ExpandType, quantization: <CubeOption<Quantization<MP>> as CubeType>::ExpandType, config: G, ) -> <() as CubeType>::ExpandType

fn __expand_task_count( scope: &mut Scope, this: <Self as CubeType>::ExpandType, ) -> u32

impl TilingLayout for StridedTilingLayout

fn get_tile<ES: Numeric, S: StageConfig>( stage: &StageMemory<ES, Self>, x: u32, y: u32, _buffer_index: u32, ident: Ident, config: S, ) -> Tile<ES>

Returns the tile at shared memory coordinates

fn __expand_get_tile<ES: Numeric, S: StageConfig>( scope: &mut Scope, stage: <StageMemory<ES, Self> as CubeType>::ExpandType, x: <u32 as CubeType>::ExpandType, y: <u32 as CubeType>::ExpandType, _buffer_index: u32, ident: Ident, config: S, ) -> <Tile<ES> as CubeType>::ExpandType

impl Copy for StridedTilingLayout