A run configuration class for running operators conforming to the standard operator signature.

Public members¶

store_output : bool = True

Whether to store the output to zarr or not.

input_obj : Input

The input object to use.

input_version : Version

The input version to use.

cluster_config : ClusterConfig = ClusterConfig(workers_per_job=1, cores_per_worker=1, omp_parallelism=False, exclusive_jobs=False, queuing=False)

The cluster configuration to use

workers : int = 1

The number of dask workers to spawn

create_report : bool = False

Whether to create a dask report.

impl : Callable

The operator implementation to run

iterations : int = 5

The number of iterations to run. The runtimes will be combined according to runtime_reduction.

warmup : bool = True

If set to True, an additional warmup iteration will be added at the start of the measurement iterations, whose runtime will be discarded.

Methods¶

load_input() → list[Any]

Loads the input for the implementation.

construct_output(*args: Dataset) → list[dask.typing.DaskCollection]

Abstract class which constructs the output dask collections to be computed.

measure() → DaskRuntime

Measures the runtime of the implementation.

setup() → None

Sets up the environment for this configuration. This will be called once before the measurement iterations start.

cleanup() → None

Perform cleanup after the measurement iterations.

runtime_reduction(axis=None, dtype=None, out=None, ...)

Compute the arithmetic mean along the specified axis, ignoring NaNs.

classmethod list_configs(**kwargs: Any) → list

Returns a list of run configurations, which is the euclidean product between the given lists of individual configurations.

Constructors¶

OperatorRunConfig(...)

Initialize self. See help(type(self)) for accurate signature.

String representation¶

__repr__()

Return repr(self).

Special methods¶

__eq__(other)

Return self==value.