P1 Smartcity: GPU Metrics Utilization

It is asked for the 2nd Phase of CODECO to integrate the following metrics - in the same way that we are using CPU metrics:

GPU Node Usage

• record: instance:gpu_usage:avg1m

   expr: avg by (instance) (rate(jetson_prometheus_exporter_gpu{statistic="val"}[1m])) 

• record: instance:gpu_frequency:avg1m

   expr: avg by (instance) (jetson_prometheus_exporter_gpu{statistic="freq"})

CODECO U1 - Rate of items processed / Throughput

• record: codeco_u1_rate

  expr: sum by(node, source, type) (rate(codeco_pointpillars_items_total[$__rate_interval]))

@yanlong01