#Accelerator Monitoring

Accelerator monitoring in ACP connects product-provided metrics with ACP observability features.

#What you can monitor

The exact metric names and dashboards depend on the accelerator product installed in the cluster. ACP observability pages explain how to view dashboards, query metrics, and configure alerts.

#Metric sources

• NVIDIA GPU metrics may come from DCGM Exporter or product-specific collectors.
• HAMi may provide virtual GPU allocation and usage metrics.
• Ascend NPU products may provide NPU exporter metrics.

#Related pages

• Use ACP Observability documentation for monitoring concepts and dashboard management.
• Use Alauda Cost Management documentation for GPU or NPU billing models.
• Use each accelerator product site for product-specific metric names and dashboards.