Structs§
- Buffer
Handle - Backend-owned buffer handle.
- Device
Id - Concrete compute device identifier.
- DynRank
- Dynamic tensor rank marker.
- Placement
- Placement metadata for a tensor buffer.
- Rank
- Static tensor rank marker.
- Strided
Slice Spec - One-axis slice specification for typed view slicing.
- Tensor
Layout - Storage-neutral tensor layout metadata.
- Tensor
Owned View - Owned lazy tensor view over a shared base tensor.
- Typed
Tensor - Runtime typed tensor storage with compile-time scalar type and rank metadata.
- Typed
Tensor View - Read-only borrowed view of typed tensor storage with arbitrary strides.
- Typed
Tensor View Mut - Mutable borrowed view of typed tensor storage with arbitrary strides.
Enums§
- Buffer
- Tensor storage.
- DType
- Runtime scalar dtype tag.
- Device
Kind - Compute device family.
- GpuBackend
Kind - GPU backend family used by placement metadata.
- Memory
Kind - Memory location for tensor storage.
- Tensor
- Dynamic tensor enum over the supported scalar types.
- Tensor
Buffer Ref - Borrowed tensor buffer reference used by read-only typed views.
- Tensor
Buffer RefMut - Borrowed tensor buffer reference used by mutable typed views.
- Tensor
Read - Read-only tensor input accepted by synchronous eager kernels.
- Tensor
Value - Owned tensor value that can be compact or a lazy view.
- Tensor
View - Dynamic read-only borrowed tensor view.
Traits§
- Backend
Buffer - Opaque backend-owned tensor buffer.
- Tensor
Rank - Rank contract for tensor metadata shapes and strides.
- Tensor
Scalar - Sealed trait for scalar types that can be stored in a
Tensor.
Functions§
- col_
major_ strides - Column-major strides derived from a shape.