Enum Output 

pub enum Output<'a, E: Numeric> {
    Tensor(TileArg<'a>),
    Atomic(AccumulateArg<'a, E>),
}

The operand a kernel leaves by: a tensor read and replaced (TileArg), or an atomic tensor added into and never read (AccumulateArg).

Variants

Tensor(TileArg<'a>)

Atomic(AccumulateArg<'a, E>)

Implementations

impl<'a, E: Numeric> Output<'a, E>

pub fn new_Tensor(_0: TileArg<'a>) -> Self

pub fn __expand_new_Tensor( _: &Scope, _0: <TileArg<'a> as CubeType>::ExpandType, ) -> OutputExpand<'a, E>

pub fn new_Atomic(_0: AccumulateArg<'a, E>) -> Self

pub fn __expand_new_Atomic( _: &Scope, _0: <AccumulateArg<'a, E> as CubeType>::ExpandType, ) -> OutputExpand<'a, E>

impl<'a, E: Numeric> Output<'a, E>

pub fn tile(&self, partitioning: &Partitioning) -> Tile

Serve this operand as a Tile under the kernel’s one partitioning.

pub fn __expand_tile( scope: &Scope, this: &<Self as CubeType>::ExpandType, partitioning: &<Partitioning as CubeType>::ExpandType, ) -> <Tile as CubeType>::ExpandType

Serve this operand as a Tile under the kernel’s one partitioning.

Trait Implementations

impl<'a, E: Numeric> CubeDebug for Output<'a, E>

fn set_debug_name(&self, scope: &Scope, name: &'static str)

Set the debug name of this type’s expansion. Should do nothing for types that don’t appear at runtime

impl<'a, E: Numeric> CubeType for Output<'a, E>

type ExpandType = OutputExpand<'a, E>

impl<'a, E: Numeric> LaunchArg for Output<'a, E>

where AccumulateArg<'a, E>: LaunchArg,

type RuntimeArg = OutputArgs<'a, E>

The runtime argument for the kernel.

type CompilationArg = OutputCompilationArg<'a, E>

Compilation argument.

fn register( runtime_arg: Self::RuntimeArg, launcher: &mut KernelLauncher, ) -> Self::CompilationArg

fn expand( arg: &Self::CompilationArg, builder: &mut KernelBuilder, ) -> <Self as CubeType>::ExpandType

Register a variable during compilation that fill the KernelBuilder.