OverlapBudgetScenario
Communication scenarios · Implementation: mlsysim/engine/scenario_definitions/communication.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
Layer-by-layer gradient AllReduce of workload: sequential against overlapped with backward.
Notes
Equation: Compose mlsysim.solvers.OverlapBudgetModel.
Assumptions: The stated fields are inputs to mlsysim.solvers.OverlapBudgetModel; its documented validity regime, units, and modeling choices apply unchanged.
Excludes: This scenario adds no measurement, empirical calibration, or mechanisms beyond mlsysim.solvers.OverlapBudgetModel.
References
Derivation: this scenario binds its input fields to the referenced implementation contracts: mlsysim.solvers.OverlapBudgetModel. The methods below identify any additional arithmetic.
Examples
>>> from mlsysim import Scenarios
>>> case = Scenarios.OverlapBudget7B64Gpu
>>> type(case.result).__name__
'OverlapBudgetResult'Attributes
| Name | Description |
|---|---|
| backward_per_layer | |
| bucket_size | |
| bytes_per_value | |
| n_devices | |
| result | Evaluate this scenario through mlsysim.solvers.OverlapBudgetModel.solve. Equation: |
| time_per_bucket | |
| workload |
workload
Type: Workload.
Declared default: required.
bytes_per_value
Type: Quantity.
Declared default: required.
backward_per_layer
Type: Quantity.
Declared default: required.
bucket_size
Type: Quantity.
Declared default: required.
time_per_bucket
Type: Quantity.
Declared default: required.
n_devices
Type: int.
Declared default: required.
result
engine.scenario_types.OverlapBudgetScenario.result
Evaluate this scenario through mlsysim.solvers.OverlapBudgetModel.solve. Equation: pass the stated fields to that implementation; its contract defines the calculation and feasibility tests. Units: each returned field follows the referenced result schema; no independent equation is evaluated by this wrapper.