EnergyPerInferenceScenario
Sustainability scenarios · Implementation: mlsysim/engine/scenario_definitions/sustainability.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
Energy per inference across tiers: solver-derived device energies plus stated anchors.
Notes
Equation: Use the dimensional products, ratios, or selection rules documented on the calculated properties below.
Assumptions: Named modeled runs and external energy anchors retain their distinct provenance; a modeled run replaces a duplicate anchor name.
Excludes: Ratios do not establish equal task quality or identical system boundaries across entries.
References
Derivation: the calculated properties below state their defining arithmetic and units. They are scenario accounting or input-selection rules, not independently measured relationships.
Examples
>>> from mlsysim import Scenarios
>>> case = Scenarios.EnergyPerInferenceTiers
>>> print(f"{next(iter(case.energies.values())).m_as('microjoule'):.1f} microjoule")
10.0 microjouleAttributes
| Name | Description |
|---|---|
| anchors | |
| energies | Named per-inference energies combining supplied anchors and modeled runs. Equation: copy |
| runs |
Methods
| Name | Description |
|---|---|
| span | Energy ratio between two named entries. Equation: energies[high] / energies[low]. |
span
engine.scenario_types.EnergyPerInferenceScenario.span(low, high)Energy ratio between two named entries. Equation: energies[high] / energies[low]. Units: dimensionless; comparisons require equivalent task definitions.
anchors
Type: Dict[str, Quantity].
Declared default: required.
runs
Type: Dict[str, SingleNodeInferenceScenario].
Declared default: required.
energies
engine.scenario_types.EnergyPerInferenceScenario.energies
Named per-inference energies combining supplied anchors and modeled runs. Equation: copy anchors then replace matching keys with run.result.energy. Units: joules for modeled runs; anchors retain their supplied units and provenance.