local

package
v0.7.3 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Feb 11, 2024 License: Apache-2.0 Imports: 13 Imported by: 0

Documentation

Index

Constants

This section is empty.

Variables

This section is empty.

Functions

This section is empty.

Types

type AllWeights 

type AllWeights struct {
	BiasWeight           float64                                  `json:"Bias_Weight"`
	CategoricalVariables map[string]map[string]CategoricalFeature `json:"Categorical_Variables"`
	NumericalVariables   map[string]NormalizedNumericalFeature    `json:"Numerical_Variables"`
}

type CategoricalFeature 

type CategoricalFeature struct {
	Weight float64 `json:"weight"`
}

type ComponentFeatures added in v0.5.4 

type ComponentFeatures int

The ratio features list follows this order 

const (
	PkgUsageMetric ComponentFeatures = iota
	CoreUsageMetric
	DramUsageMetric
	UncoreUsageMetric
	OtherUsageMetric
	GpuUsageMetric
	PkgDynPower
	CoreDynPower
	DramDynPower
	UncoreDynPower
	OtherDynPower
	GpuDynPower
	PkgIdlePower
	CoreIdlePower
	DramIdlePower
	UncoreIdlePower
	OtherIdlePower
	GpuIdlePower
)

The ratio power model can be used for both node Component and Platform power estimation. The models for Component and Platform power estimation have different number of features.

type ComponentModelWeights 

type ComponentModelWeights map[string]ModelWeights

ComponentModelWeights defines structure for multiple (power component's) weights { "core": 

{"All_Weights":
  {
  "Bias_Weight": 1.0,
  "Categorical_Variables": {"cpu_architecture": {"Sky Lake": {"weight": 1.0}}},
  "Numerical_Variables": {"cpu_cycles": {"mean": 0, "variance": 1.0, "weight": 1.0}}
  }
},

"dram": {"All_Weights": 

  {
  "Bias_Weight": 1.0,
  "Categorical_Variables": {"cpu_architecture": {"Sky Lake": {"weight": 1.0}}},
  "Numerical_Variables": {"cache_miss": {"mean": 0, "variance": 1.0, "weight": 1.0}}
  }
}

type LinearRegressor 

type LinearRegressor struct {
	ModelServerEndpoint  string
	OutputType           types.ModelOutputType
	EnergySource         string
	TrainerName          string
	SelectFilter         string
	ModelWeightsURL      string
	ModelWeightsFilepath string

	FloatFeatureNames           []string
	SystemMetaDataFeatureNames  []string
	SystemMetaDataFeatureValues []string
	// contains filtered or unexported fields
}

LinearRegressor defines power estimator with linear regression approach

func (*LinearRegressor) AddDesiredOutValue added in v0.5.4 

func (r *LinearRegressor) AddDesiredOutValue(y float64)

AddDesiredOutValue adds the the y, which is the response variable (or the dependent variable) of regression. LinearRegressor is trained off-line then we do not add Y for trainning. We might implement it in the future.

func (*LinearRegressor) AddNodeFeatureValues added in v0.5.4 

func (r *LinearRegressor) AddNodeFeatureValues(x []float64)

AddNodeFeatureValues adds the the x for prediction, which is the variable used to calculate the ratio. LinearRegressor is not trained, then we cannot Add training samples, only samples for prediction. The LinearRegressor does not differentiate node or process power estimation, the difference will only be the amount of resource utilization

func (*LinearRegressor) AddProcessFeatureValues added in v0.7.3 

func (r *LinearRegressor) AddProcessFeatureValues(x []float64)

AddProcessFeatureValues adds the the x for prediction, which are the explanatory variables (or the independent variable) of regression. LinearRegressor is trained off-line then we cannot Add training samples. We might implement it in the future. The LinearRegressor does not differentiate node or process power estimation, the difference will only be the amount of resource utilization

func (*LinearRegressor) GetComponentsPower added in v0.5.4 

func (r *LinearRegressor) GetComponentsPower(isIdlePower bool) ([]source.NodeComponentsEnergy, error)

GetComponentsPower applies each component's ModelWeight prediction and return a map of component power associated to each process/process/pod

func (*LinearRegressor) GetGPUPower added in v0.5.4 

func (r *LinearRegressor) GetGPUPower(isIdlePower bool) ([]float64, error)

GetComponentsPower returns GPU Power in Watts associated to each each process

func (*LinearRegressor) GetModelType added in v0.5.4 

func (r *LinearRegressor) GetModelType() types.ModelType

GetModelType returns the model type

func (*LinearRegressor) GetNodeFeatureNamesList added in v0.5.4 

func (r *LinearRegressor) GetNodeFeatureNamesList() []string

GetNodeFeatureNamesList returns the list of float features that the model was configured to use The LinearRegressor does not differentiate node or process power estimation, the difference will only be the amount of resource utilization

func (*LinearRegressor) GetPlatformPower added in v0.5.4 

func (r *LinearRegressor) GetPlatformPower(isIdlePower bool) ([]float64, error)

GetPlatformPower applies ModelWeight prediction and return a list of power associated to each process/process/pod

func (*LinearRegressor) GetProcessFeatureNamesList added in v0.7.3 

func (r *LinearRegressor) GetProcessFeatureNamesList() []string

GetProcessFeatureNamesList returns the list of float features that the model was configured to use The LinearRegressor does not differentiate node or process power estimation, the difference will only be the amount of resource utilization

func (*LinearRegressor) IsEnabled added in v0.5.4 

func (r *LinearRegressor) IsEnabled() bool

IsEnabled returns true if the power model was trained and is active

func (*LinearRegressor) ResetSampleIdx added in v0.5.4 

func (r *LinearRegressor) ResetSampleIdx()

ResetSampleIdx set the sample vector index to 0 to overwrite the old samples with new ones for trainning or prediction.

func (*LinearRegressor) Start added in v0.5.4 

func (r *LinearRegressor) Start() error

Start returns nil if model weight is obtainable

func (*LinearRegressor) Train added in v0.5.4 

func (r *LinearRegressor) Train() error

Train triggers the regressiong fit after adding data points to create a new power model. LinearRegressor is trained off-line then we cannot trigger the trainning. We might implement it in the future.

type ModelRequest 

type ModelRequest struct {
	MetricNames  []string `json:"metrics"`
	OutputType   string   `json:"output_type"`
	EnergySource string   `json:"source"`
	NodeType     int      `json:"node_type"`
	Weight       bool     `json:"weight"`
	TrainerName  string   `json:"trainer_name"`
	SelectFilter string   `json:"filter"`
}

ModelRequest defines a request to Kepler Model Server to get model weights

type ModelWeights 

type ModelWeights struct {
	AllWeights `json:"All_Weights"`
}

ModelWeights, AllWeight, CategoricalFeature, NormalizedNumericalFeature define structure of model weight { "All_Weights": 

	{
	"Bias_Weight": 1.0,
	"Categorical_Variables": {"cpu_architecture": {"Sky Lake": {"weight": 1.0}}},
	"Numerical_Variables": {"cpu_cycles": {"mean": 0, "variance": 1.0, "weight": 1.0}}
	}
}

type NormalizedNumericalFeature 

type NormalizedNumericalFeature struct {
	Scale  float64 `json:"scale"` // to normalize the data
	Weight float64 `json:"weight"`
}

type PlaformFeatures added in v0.5.4 

type PlaformFeatures int
const (
	// Node features are at the end of the feature list
	PlatformUsageMetric PlaformFeatures = iota
	PlatformDynPower
	PlatformIdlePower
)

type RatioPowerModel added in v0.5.4 

type RatioPowerModel struct {
	NodeFeatureNames    []string
	ProcessFeatureNames []string
	// contains filtered or unexported fields
}

RatioPowerModel stores the feature samples of each process for prediction The feature will added ordered from cycles, instructions and cache-misses

func (*RatioPowerModel) AddDesiredOutValue added in v0.5.4 

func (r *RatioPowerModel) AddDesiredOutValue(y float64)

AddDesiredOutValue adds the the y, which is the response variable (or the dependent variable) of regression. RatioPowerModel is not trained, then we cannot Add training samples.

func (*RatioPowerModel) AddNodeFeatureValues added in v0.5.4 

func (r *RatioPowerModel) AddNodeFeatureValues(x []float64)

AddNodeFeatureValues adds the the x for prediction, which is the variable used to calculate the ratio. RatioPowerModel is not trained, then we cannot Add training samples, only samples for prediction.

func (*RatioPowerModel) AddProcessFeatureValues added in v0.7.3 

func (r *RatioPowerModel) AddProcessFeatureValues(x []float64)

AddProcessFeatureValues adds the the x for prediction, which is the variable used to calculate the ratio. RatioPowerModel is not trained, then we cannot Add training samples, only samples for prediction.

func (*RatioPowerModel) GetComponentsPower added in v0.5.4 

func (r *RatioPowerModel) GetComponentsPower(isIdlePower bool) ([]source.NodeComponentsEnergy, error)

GetComponentsPower applies each component's ModelWeight prediction and return a map of component powers

func (*RatioPowerModel) GetGPUPower added in v0.5.4 

func (r *RatioPowerModel) GetGPUPower(isIdlePower bool) ([]float64, error)

GetComponentsPower returns GPU Power in Watts associated to each each process/process/pod

func (*RatioPowerModel) GetModelType added in v0.5.4 

func (r *RatioPowerModel) GetModelType() types.ModelType

GetModelType returns the model type

func (*RatioPowerModel) GetNodeFeatureNamesList added in v0.5.4 

func (r *RatioPowerModel) GetNodeFeatureNamesList() []string

GetNodeFeatureNamesList returns the list of process features that the model was configured to use

func (*RatioPowerModel) GetPlatformPower added in v0.5.4 

func (r *RatioPowerModel) GetPlatformPower(isIdlePower bool) ([]float64, error)

GetPlatformPower applies ModelWeight prediction and return a list of total powers

func (*RatioPowerModel) GetProcessFeatureNamesList added in v0.7.3 

func (r *RatioPowerModel) GetProcessFeatureNamesList() []string

GetProcessFeatureNamesList returns the list of process features that the model was configured to use

func (*RatioPowerModel) IsEnabled added in v0.5.4 

func (r *RatioPowerModel) IsEnabled() bool

IsEnabled returns true as Ratio Power model is always active

func (*RatioPowerModel) ResetSampleIdx added in v0.5.4 

func (r *RatioPowerModel) ResetSampleIdx()

ResetSampleIdx set the sample vector index to 0 to overwrite the old samples with new ones for trainning or prediction.

func (*RatioPowerModel) Train added in v0.5.4 

func (r *RatioPowerModel) Train() error

RatioPowerModel is not trained, then this function does nothing.

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.