SustainedThroughputScenario

Training scenarios · Implementation: mlsysim/engine/scenario_definitions/training.py.

Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.

A run budgeted from peak throughput, then run at its sustained share of peak.

Device-hours scale as 1 / sustained_fraction: a plan made from the datasheet rate needs that many times the hours when the workload sustains only part of peak.

Notes

Equation: Use the dimensional products, ratios, or selection rules documented on the calculated properties below.

Assumptions: The work budget is fixed and the positive sustained fraction scales delivered arithmetic capacity uniformly.

Excludes: Changes to algorithmic work and time-varying utilization are absent.

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.PeakPlannedTrainingBudget
>>> round(case.required_device_hours, 4)
2500.0

Attributes

Name	Description
planned_device_hours
required_device_hours	Device-time needed when only a fraction of peak is delivered. Equation:
stretch	Run-time multiplier relative to the peak-rate plan. Equation: ``1 /
sustained_fraction

planned_device_hours

Type: float.

Declared default: required.

sustained_fraction

Type: float.

Declared default: required.

required_device_hours

engine.scenario_types.SustainedThroughputScenario.required_device_hours

Device-time needed when only a fraction of peak is delivered. Equation: planned_device_hours / sustained_fraction. Units: device-hours; the fraction must be positive.

stretch

engine.scenario_types.SustainedThroughputScenario.stretch

Run-time multiplier relative to the peak-rate plan. Equation: 1 / sustained_fraction. Units: dimensionless; total work is fixed.