Enum DTypeUsage 

pub enum DTypeUsage {
    Storage,
    Arithmetic,
    Accelerated,
}

Describes how a data type can be used on a given device.

A data type may be supported for different classes of operations. Not all data types that appear in hardware or kernel implementations are suitable for general-purpose tensor operations.

Variants

Storage

The type can be stored in device memory and converted to and from other supported data types.

Arithmetic

The type supports general-purpose arithmetic and common tensor operations (e.g. elementwise ops, reductions, etc.).

Accelerated

The type is supported by hardware-accelerated execution paths.

This typically indicates support for accelerator-backed compute units (e.g., tensor cores executing MMA instructions) for high-performance operations such as matrix multiplication and operations that lower to it.

Notes

• A type can be both Arithmetic and Accelerated if it supports general-purpose operations and accelerated paths.
• If a type is marked as Accelerated but not Arithmetic, it is not suitable for general-purpose tensor operations and may only be used in specific accelerated operations.

Accelerated is a flag, not a detailed descriptor. It does not enumerate which operations are accelerated or which accelerator features are available.

Implementations

impl DTypeUsage

pub fn general() -> EnumSet<DTypeUsage>

Returns the usage set required for general-purpose tensor support.

Trait Implementations

impl<O> BitAnd<O> for DTypeUsage

where O: Into<EnumSet<DTypeUsage>>,

type Output = EnumSet<DTypeUsage>

The resulting type after applying the & operator.

fn bitand(self, other: O) -> <DTypeUsage as BitAnd<O>>::Output

Performs the & operation.

impl<O> BitOr<O> for DTypeUsage

where O: Into<EnumSet<DTypeUsage>>,

type Output = EnumSet<DTypeUsage>

The resulting type after applying the | operator.

fn bitor(self, other: O) -> <DTypeUsage as BitOr<O>>::Output

Performs the | operation.

impl<O> BitXor<O> for DTypeUsage

where O: Into<EnumSet<DTypeUsage>>,

type Output = EnumSet<DTypeUsage>

The resulting type after applying the ^ operator.

fn bitxor(self, other: O) -> <DTypeUsage as BitXor<O>>::Output

Performs the ^ operation.

impl Clone for DTypeUsage

fn clone(&self) -> DTypeUsage

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for DTypeUsage

impl Debug for DTypeUsage

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

Formats the value using the given formatter.

impl EnumSetType for DTypeUsage

impl Eq for DTypeUsage

impl Not for DTypeUsage

type Output = EnumSet<DTypeUsage>

The resulting type after applying the ! operator.

fn not(self) -> <DTypeUsage as Not>::Output

Performs the unary ! operation.

impl PartialEq for DTypeUsage

fn eq(&self, other: &DTypeUsage) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl PartialEq<EnumSet<DTypeUsage>> for DTypeUsage

fn eq(&self, other: &EnumSet<DTypeUsage>) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl<O> Sub<O> for DTypeUsage

where O: Into<EnumSet<DTypeUsage>>,

type Output = EnumSet<DTypeUsage>

The resulting type after applying the - operator.

fn sub(self, other: O) -> <DTypeUsage as Sub<O>>::Output

Performs the - operation.