pygwtf.backend#

Attributes#

gpu_available

Classes#

`Backend`	Backend base class. Defines the interface for the array module to be used for computations (e.g. numpy or cupy)
`CPUBackend`	Backend base class. Defines the interface for the array module to be used for computations (e.g. numpy or cupy)
`GPUBackend`	Backend base class. Defines the interface for the array module to be used for computations (e.g. numpy or cupy)

Functions#

get_backend(→ Backend)

Factory function to get the appropriate backend instance based on the provided backend name.

Module Contents#

pygwtf.backend.gpu_available = True#

class pygwtf.backend.Backend#

Bases: abc.ABC

Backend base class. Defines the interface for the array module to be used for computations (e.g. numpy or cupy) and any other backend-specific functionality (such as kernel selection).

property xp: Any#

Abstractmethod:

Returns the array module to be used for computations. This will be numpy for CPU computations and cupy for GPU computations.

property uses_gpu: bool#: Returns True if the backend uses GPU computations, False otherwise.

asnumpy(array: Any) → numpy.ndarray#: Converts the given array to a NumPy array. For CPU backend, this can be a no-op. For GPU backend, this will transfer the array from GPU to CPU.

class pygwtf.backend.CPUBackend#

Bases: Backend

Backend base class. Defines the interface for the array module to be used for computations (e.g. numpy or cupy) and any other backend-specific functionality (such as kernel selection).

property xp#: Returns the array module to be used for computations. This will be numpy for CPU computations and cupy for GPU computations.

class pygwtf.backend.GPUBackend#

Bases: Backend

Backend base class. Defines the interface for the array module to be used for computations (e.g. numpy or cupy) and any other backend-specific functionality (such as kernel selection).

property xp#: Returns the array module to be used for computations. This will be numpy for CPU computations and cupy for GPU computations.

property uses_gpu: bool#: Returns True if the backend uses GPU computations, False otherwise.

asnumpy(array: Any) → numpy.ndarray#: Converts the given array to a NumPy array. For CPU backend, this can be a no-op. For GPU backend, this will transfer the array from GPU to CPU.

pygwtf.backend.get_backend(backend_name: str) → Backend#: Factory function to get the appropriate backend instance based on the provided backend name.