tf.config.experimental.get_memory_growth

TensorFlow 1 version View source on GitHub

Get if memory growth is enabled for a PhysicalDevice.

Aliases:

tf.config.experimental.get_memory_growth(device)

If memory growth is enabled for a PhysicalDevice, the runtime initialization will not allocate all memory on the device.

For example:

physical_devices = tf.config.list_physical_devices('GPU')
try:
tf.config.experimental.set_memory_growth(physical_devices[0], True)
assert tf.config.experimental.get_memory_growth(physical_devices[0])
except:
# Invalid device or cannot modify virtual devices once initialized
pass

Args:

  • device: PhysicalDevice to query

Returns:

A boolean indicating the memory growth setting for the PhysicalDevice.

Raises:

  • ValueError: Invalid PhysicalDevice specified.