Health

Check GPU hardware component health. Reports the status of key GPU subsystems (temperature, power, frequency) and indicates whether they are operating within normal parameters.

Note

This command is supported on Linux only.

Synopsis

xpu-smi health -l
xpu-smi health -l -j
xpu-smi health -d [deviceId]
xpu-smi health --device [deviceId]
xpu-smi health --device [pciBdfAddress]
xpu-smi health --device [deviceId] -j
xpu-smi health --device [pciBdfAddress] -j
xpu-smi health --device [deviceId] -c [componentTypeId]
xpu-smi health --device [pciBdfAddress] -c [componentTypeId] -j

Options

-h, --help

Print this help message and exit.

-j, --json

Print result in JSON format.

-l, --list

Display health information for all detected GPU devices.

-d <deviceId>, --device <deviceId>, --id <deviceId>

The device ID or PCI BDF address to query. Accepts a comma-separated list to query several devices at once (e.g., -d 0,1,4).

Note

Separate per-device output is only produced when --list is used. Without --list, results for a multi-device selection are not shown per device.

-c <componentTypeId>, --component <componentTypeId>

Filter health status to a specific component type. Valid component IDs:

ID

Component

1

GPU Core Temperature

2

GPU Memory Temperature

3

GPU Power

4

GPU Frequency

Any other value is rejected as an invalid component ID, including:

  • other numbers, such as 0, 5, 6, 999 or -1

  • letters, such as abc

  • mixed letters and digits, such as 4x or a4

  • decimals, such as 4.0

  • an empty value

Note

GPU Frequency was component ID 6 in earlier releases; it is now ID 4. Use -c 4 for GPU Frequency health. -c 6 is now rejected as invalid.

ID 4 was GPU Memory in earlier releases. GPU Memory health is no longer available: the driver reports memory health only for HBM memory, which the supported devices do not have.

Examples

List health status for all devices:

xpu-smi health -l

List health in JSON format:

xpu-smi health -l -j

Show all component health for device 0:

xpu-smi health --device 0

Show only GPU Core Temperature health for device 0:

xpu-smi health --device 0 -c 1

Show GPU frequency health for device addressed by PCI BDF:

xpu-smi health --device 0000:4d:00.0 -c 4