gpm

Documentation

Overview

Package gpm provides the NVIDIA GPM metrics collection and reporting.

Index

• Constants
• func InitAveragers(db *sql.DB, tableName string)
• func ReadGPUAnyTensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUDFMATensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUFp16UtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUFp32UtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUFp64UtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUHMMATensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUIMMATensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUIntUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func ReadGPUSMOccupancyPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)
• func Register(reg *prometheus.Registry, db *sql.DB, tableName string) error
• func SetGPUUtilPercent(ctx context.Context, metricID nvml.GpmMetricId, gpuID string, pct float64, ...) error
• func SetLastUpdateUnixSeconds(unixSeconds float64)

Constants

const SubSystem = "accelerator_nvidia_gpm"

Functions

func InitAveragers

func InitAveragers(db *sql.DB, tableName string)

func ReadGPUAnyTensorUtilPercents

func ReadGPUAnyTensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUDFMATensorUtilPercents

func ReadGPUDFMATensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUFp16UtilPercents

func ReadGPUFp16UtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUFp32UtilPercents

func ReadGPUFp32UtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUFp64UtilPercents

func ReadGPUFp64UtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUHMMATensorUtilPercents

func ReadGPUHMMATensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUIMMATensorUtilPercents

func ReadGPUIMMATensorUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUIntUtilPercents

func ReadGPUIntUtilPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func ReadGPUSMOccupancyPercents

func ReadGPUSMOccupancyPercents(ctx context.Context, since time.Time) (components_metrics_state.Metrics, error)

func Register

func Register(reg *prometheus.Registry, db *sql.DB, tableName string) error

func SetGPUUtilPercent

func SetGPUUtilPercent(ctx context.Context, metricID nvml.GpmMetricId, gpuID string, pct float64, currentTime time.Time) error

func SetLastUpdateUnixSeconds

func SetLastUpdateUnixSeconds(unixSeconds float64)