Prometheus and Grafana integration

To ingest metrics into Prometheus, add the AMD Device Metrics Exporter endpoint to your Prometheus configuration:

scrape_configs:
  - job_name: 'gpu_metrics'
    static_configs:
      - targets: ['exporter_external_ip:5000']

Pre-built Grafana dashboards are available in the grafana/ directory of the repository:

• GPU Overview Dashboard

• Per-Node Dashboard

• Job-specific Dashboard

Import these dashboards through the Grafana interface for immediate visualization of your GPU metrics.