Official TF documentation [1] suggests 2 ways to control GPU memory allocation
Memory growth allows TF to grow memory based on usage
tf.config.experimental.set_memory_growth(gpus[0], True)
Virtual device config sets limit to memory
tf.config.experimental.set_virtual_device_configuration(
gpus[0],
[tf.config.experimental.VirtualDeviceConfiguration(memory_limit=1024)])
In this case, can these two statements be used together in conjunction? or are these mutually exclusive and opposite?
Speak: Can we set memory growth to true but at the same time restrict memory limit?
Refer
[1] https://www.tensorflow.org/guide/gpu#limiting_gpu_memory_growth