|
MayaFlux 1.0.0
Digital-First Multimedia Processing Framework
|
Complete description of a tensor: type, shape, layout, meaning. More...
#include <TensorSpec.hpp>
Public Member Functions | |
| std::optional< std::pair< uint64_t, uint64_t > > | byte_range () const noexcept |
| Storage bytes [first, last) the spec addresses, from storage start. | |
| std::string | diagnose () const |
| Empty when the spec is self-consistent, otherwise the reason it is not. | |
| std::vector< DataDimension > | dimensions () const |
| One DataDimension per axis, with size and stride taken from shape and strides. | |
| std::optional< uint64_t > | element_count () const noexcept |
| Product of the shape, or std::nullopt on overflow. | |
| bool | is_contiguous () const noexcept |
| Whether elements are laid out densely in C order. | |
| bool | is_valid () const |
| Whether diagnose() reports no problem. | |
| size_t | rank () const noexcept |
| Number of dimensions. | |
| std::vector< int64_t > | resolved_strides () const |
| Element strides, with C order filled in when none are stored. | |
| std::optional< TensorSpec > | window (const Region ®ion) const |
| Spec of the box a Region selects. | |
| std::optional< TensorSpec > | window (std::span< const uint64_t > origin, std::span< const uint64_t > extent) const |
Spec of the box that starts at origin and spans extent elements per axis. | |
Static Public Member Functions | |
| static TensorSpec | contiguous (TensorDType dtype, std::vector< uint64_t > shape, std::string name={}) |
| C-ordered spec with no offset, axes or quantization. | |
Public Attributes | |
| std::vector< DataDimension > | axes |
| uint64_t | byte_offset {} |
| TensorDevice | device |
| TensorDType | dtype |
| std::string | name |
| std::optional< TensorQuantization > | quantization |
| std::vector< uint64_t > | shape |
| std::vector< int64_t > | strides |
Complete description of a tensor: type, shape, layout, meaning.
Follows the DLPack model. Strides are in elements and may be explicit, negative or broadcast (zero), so views, Fortran order and transposes are described without copying; empty strides mean C order. byte_offset locates element zero inside the storage. axes optionally gives each dimension a role and name, for example batch, channel and spatial axes.
Usage:
Definition at line 92 of file TensorSpec.hpp.