Missed TensorFlow World? Check out the recap. Learn more

tf.config.experimental.get_visible_devices

TensorFlow 2 version View source on GitHub

Get the list of visible physical devices.

Aliases:

tf.config.experimental.get_visible_devices(device_type=None)

Returns a list of PhysicalDevice objects that are current marked as visible to the runtime. Any visible devices will have LogicalDevices assigned to them once the runtime is initialized.

The following example verifies all visible GPUs have been disabled:

physical_devices = config.experimental.list_physical_devices('GPU')
assert len(physical_devices) > 0, "Not enough GPU hardware devices available"
# Disable all GPUS
tf.config.experimental.set_visible_devices([], 'GPU')
visible_devices = tf.config.experimental.get_visible_devices()
for device in visible_devices:
  assert device.device_type != 'GPU'

Args:

  • device_type: (optional) Device types to limit query to.

Returns:

List of PhysicalDevice objects