# DCGM Metrics

CM for metrics

apiVersion: v1
kind: ConfigMap
metadata:
  name: dcgm-exporter-metrics
  namespace: nvidia-gpu-operator
data:
  dcgm-metrics.csv: |
    # Essential metrics for workload attribution
    DCGM_FI_PROF_GR_ENGINE_ACTIVE, gauge, Graphics engine active time ratio
    DCGM_FI_DEV_MEM_COPY_UTIL, gauge, Memory utilization percentage
    DCGM_FI_DEV_GPU_TEMP, gauge, GPU temperature (in C)
    DCGM_FI_DEV_POWER_USAGE, gauge, Power draw (in W)
    DCGM_FI_DEV_FB_USED, gauge, Frame buffer memory used (MiB)
    DCGM_FI_DEV_FB_FREE, gauge, Frame buffer memory free (MiB)
    DCGM_FI_DEV_SM_CLOCK, gauge, SM clock frequency (MHz)
    DCGM_FI_DEV_MEM_CLOCK, gauge, Memory clock frequency (MHz)
    DCGM_FI_PROF_SM_ACTIVE, gauge, SM active time ratio
    DCGM_FI_PROF_SM_OCCUPANCY, gauge, SM occupancy (for MIG instances)
    DCGM_FI_PROF_PIPE_TENSOR_ACTIVE, gauge, Tensor pipe utilization

DCGM_FI_PROF_GR_ENGINE_ACTIVE
DCGM_FI_DEV_MEM_COPY_UTIL
DCGM_FI_DEV_GPU_TEMP
DCGM_FI_DEV_POWER_USAGE
DCGM_FI_DEV_FB_USED
DCGM_FI_DEV_FB_FREE
DCGM_FI_DEV_SM_CLOCK
DCGM_FI_DEV_MEM_CLOCK
DCGM_FI_PROF_SM_ACTIVE
DCGM_FI_PROF_SM_OCCUPANCY
DCGM_FI_PROF_PIPE_TENSOR_ACTIVE

service monitor