Tensorflow 1.14 specify GPU running settings

Be careful ：allow_soft_placement=True indicate ： Computing devices are optional , If you don't have this parameter , Will report a mistake .
Because not all operations can be placed in GPU On , If forced will not be able to put in GPU The operation on is assigned to GPU On , Will be an error .

os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = "0"    #  Here is gpu The serial number of , Designated to use gpu object 
gpu_options = tf.GPUOptions()
gpu_options.allow_growth = True
config_photo = tf.ConfigProto(allow_soft_placement=True, gpu_options=gpu_options, log_device_placement=True)
self.sess = tf.Session(config=config_photo)