tf.config.list_physical_devices

Return a list of physical devices visible to the host runtime.

View aliases

Main aliases

tf.config.experimental.list_physical_devices

Compat aliases for migration

See Migration guide for more details.

tf.compat.v1.config.experimental.list_physical_devices, tf.compat.v1.config.list_physical_devices

tf.config.list_physical_devices(
    device_type=None
)

Used in the notebooks

Used in the guide	Used in the tutorials
Use a GPU TensorFlow basics Distributed training with Core APIs and DTensor Distributed training with TensorFlow DTensor concepts	Customization basics: tensors and operations Custom layers Using DTensors with Keras Distributed training with DTensors Distributed Input

Physical devices are hardware devices present on the host machine. By default all discovered CPU and GPU devices are considered visible.

This API allows querying the physical hardware resources prior to runtime initialization. Thus, giving an opportunity to call any additional configuration APIs. This is in contrast to tf.config.list_logical_devices, which triggers runtime initialization in order to list the configured devices.

The following example lists the number of visible GPUs on the host.

physical_devices = tf.config.list_physical_devices('GPU')
print("Num GPUs:", len(physical_devices))
Num GPUs: ...

However, the number of GPUs available to the runtime may change during runtime initialization due to marking certain devices as not visible or configuring multiple logical devices.

Args
device_type	(optional string) Only include devices matching this device type. For example "CPU" or "GPU".
Notes	If provided with any numerical values or any string other than supported device type such as 'CPU' it returns an empty list instead of raising error. 2. For default value it returns all physical devices

Returns
List of discovered tf.config.PhysicalDevice objects