hysop.backend.device.opencl.operator.transpose module

class hysop.backend.device.opencl.operator.transpose.OpenClTranspose(input_field, output_field, variables, axes, name=None, pretty_name=None, **kwds)

Bases: TransposeOperatorBase, OpenClOperator

Initialize a transposition operator operating on CartesianTopologyDescriptors.

input_field: ScalarField

Input continuous scalar field to be transposed, at least 2D.

output_field: ScalarField

Output continuous scalar field where the result is stored Transposed shape should match the input. output_field can be the same as input_field resulting in an inplace transposition.

variables: dict

Dictionary of fields as keys and CartesianTopologyDescriptors as values. Should contain input and output field.

axes: tuple or list of ints

Permutation of axes in numpy notations Axe dim-1 is the contiguous axe, axe 0 has the greatest stride in memory.

kwds: dict

Base class keyword arguments.

apply(**kwds)

Abstract method that should be implemented. Applies this node (operator, computational graph operator…).

enqueue_copy_kernel(_dst, _src, queue)

setup(work)

Setup temporary buffer that have been requested in get_work_properties(). This function may be used to execute post allocation routines. This sets self.ready flag to True. Once this flag is set one may call ComputationalGraphNode.apply() and ComputationalGraphNode.finalize().

Automatically honour temporary field memory requests.

classmethod supports_mpi()

Return True if this operator was implemented to support multiple mpi processes.