Skip to main content

Overview panel

The Overview panel is a summary assembled from the same metrics shown in the detailed panels. It does not collect separate data.

Headline charts

ChartMetric/calculationMeaning
Thermal MaxMaximum finite temperature across all current thermal metricsHottest reported on-die or board sensor at each cache sample. The sensor producing the maximum can change over time.
Current Powerpower_current_wattsLatest valid total of PMBus rails successfully read in the corresponding power pass. See Power for partial-read behavior.
CPUcpu_usage_pctAggregate Linux CPU busy percentage over the latest system sample window.
Memorylinux_mem_used_pct(MemTotal - MemAvailable) / MemTotal × 100.
MLA Memorymla_mem_allocated_mbAllocated bytes reported by /dev/simaai-mem, converted to MiB. This is allocator usage, not MLA utilization.
Networknet_rx_mbps + net_tx_mbpsCombined aggregate receive and transmit rate for all non-loopback interfaces. Despite the key name, the displayed unit is MiB/s-equivalent, not megabits/s.

The Thermal Max chart is useful for spotting the hottest location but should not replace the Thermal panel: it hides which sensor is responsible and can switch between on-die and board sensors.

Current status table

The table reports:

  • normalized one-minute CPU load;
  • Linux memory used in MiB;
  • MLA allocator memory in MiB;
  • current board-rail power;
  • eMMC and optional NVMe filesystem usage;
  • aggregate receive and transmit rates.

The columns mean:

  • Metric: short display name from the cache metric definition.
  • Group: subsystem category.
  • Value: latest finite cached value.
  • Status: normal, warning, critical, or unknown, based on the metric's configured thresholds.
  • Trend: normalized sparkline across retained cache history.

Notes panel

The Notes panel displays top-level daemon errors when present. With no daemon errors it displays usage guidance. An empty Notes panel is not proof that every sensor is healthy; Power has its own degraded status and individual metric values can be unavailable without becoming top-level daemon errors.