RTX显卡 运行TensorFlow=1.14.0 代码 报错 Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR

硬件环境:

RTX2070super 显卡

软件环境:

Ubuntu18.04.5 

 Tensorflow = 1.14.0

 ---------------------------------------------------------------------

运行代码:

import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)

def dense(x, size, scope):
    return tf.contrib.layers.fully_connected(x, size, 
                        activation_fn=None, scope=scope)

def dense_relu(x, size, scope):
    with tf.variable_scope(scope):
        h1 = dense(x, size, 'dense')
        return tf.nn.relu(h1, 'relu')


tf.reset_default_graph()
x = tf.placeholder('float32', (None, 784), name='x')
y = tf.placeholder('float32', (None, 10), name='y')
phase = tf.placeholder(tf.bool, name='phase')

h1 = dense_relu(x, 100, 'layer1')
h1 = tf.contrib.layers.batch_norm(h1, 
                            center=True, scale=True, 
                            is_training=phase,
                            scope='bn_1')

h2 = dense_relu(h1, 100, 'layer2')
h2 = tf.contrib.layers.batch_norm(h2, 
                            center=True, scale=True, 
                            is_training=phase,
                            scope='bn_2')

logits = dense(h2, 10, scope='logits')

with tf.name_scope('accuracy'):
    accuracy = tf.reduce_mean(tf.cast(
            tf.equal(tf.argmax(y, 1), tf.argmax(logits, 1)), 
            'float32'))

with tf.name_scope('loss'):
    loss = tf.reduce_mean(
        tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=y))


def train():
    update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
    with tf.control_dependencies(update_ops):
        train_step = tf.train.GradientDescentOptimizer(0.01).minimize(loss)
    sess = tf.Session()
    sess.run(tf.global_variables_initializer())
    
    history = []
    iterep = 500
    for i in range(iterep * 30):
        x_train, y_train = mnist.train.next_batch(100)
        sess.run(train_step,
                 feed_dict={'x:0': x_train, 
                            'y:0': y_train, 
                            'phase:0': 1})
        if (i + 1) %  iterep == 0:
            epoch = (i + 1)/iterep
            tr = sess.run([loss, accuracy], 
                          feed_dict={'x:0': mnist.train.images,
                                     'y:0': mnist.train.labels,
                                     'phase:0': 1})
            t = sess.run([loss, accuracy], 
                         feed_dict={'x:0': mnist.test.images,
                                    'y:0': mnist.test.labels,
                                    'phase:0': 0})
            history += [[epoch] + tr + t]
            print( history[-1] )
    return history


train()

报错,   具体如下:

2020-08-09 21:03:53.837785: E tensorflow/stream_executor/cuda/cuda_dnn.cc:329] Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR
2020-08-09 21:03:53.837987: W ./tensorflow/stream_executor/stream.h:1995] attempting to perform DNN operation using StreamExecutor without DNN support
Traceback (most recent call last):
  File "/home/devil/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1356, in _do_call
    return fn(*args)
  File "/home/devil/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1341, in _run_fn
    options, feed_dict, fetch_list, target_list, run_metadata)
  File "/home/devil/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1429, in _call_tf_sessionrun
    run_metadata)
tensorflow.python.framework.errors_impl.InternalError: cuDNN launch failure : input shape ([100,100,1,1])
     [[{{node bn_1/cond/FusedBatchNorm}}]]

During handling of the above exception, another exception occurred:

 不使用  显卡 进行计算,正常运行:

 或:

主要语句:

CUDA_VISIBLE_DEVICES=-1

正常运行:

如果  这种情况要仍然要使用 RTX 显卡, 那么 加入下面语句(对  会话session 的创建不使用默认设置,而是进行配置):

使用非交互的session时候,如下:

gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.5)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))

gpu_options = tf.GPUOptions( allow_growth = True )
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))

gpu_options = tf.GPUOptions( per_process_gpu_memory_fraction=0.5, allow_growth = True )
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))

总之,就是不能使用默认配置的session,需要配置一下。

其中, 

per_process_gpu_memory_fraction=0.5

是指为该程序分配使用的显卡其内存不超过总内存的 0.5倍。

--------------------------------------------------------

发生该问题的原因:

Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR 这个问题大部分是因为RTX显卡不兼容它出生前的接口有关。  

原因解释出自资料:

https://blog.csdn.net/pkuyjxu/article/details/89402298

-------------

对上面代码中对 tensor 进行运算的代码中  feed_dict 的形式不是很熟悉,即:

因为以前经常使用的形式为:

于是很好奇,将代码改为如下:

 发现报错:

从报错中可以知道,原来 feed_dict 中的key 是可以用 所构建的图的tensor(用函数tf.placeholder生成的tensor) 在图内的名字来表示的,即 "<op_name>:<output_index>" , 也就是这里的 “x:0” 。

而我们以前常用的形式是 将构建图中tensor (用tf.placeholder生成的tensor)的那个变量 即 x  作为 feed_dict 中的key 的。

比如:

这里,我们相当于构建了一个tensor (用函数tf.placeholder生成的tensor), tensor的名字为   'xxx:0'   , 但是所构建的这个tensor  的变量为 x 。

详细的说就是:

x = tf.placeholder('float32', (None, 784), name='x')  中, name="x" 是说这个tf.placeholer函数在图中所定义的操作( operation)的名字(name) 是 “xxx” ,  而图中的这个操作产生的第0个tensor在图中的名字为 “xxx:0”  , 而这个名字为  “xxx:0” 的tensor又传递给了python变量x , 因此在 feed_dict 中我们可以使用变量x 来表示这个tensor, 也可以使用这个tensor的图内的名字“xxx:0”  来表示。需要注意的是“xxx”是操作(operation)的名字,而不是tensor的名字。

对于 tensor 的这个   "<op_name>:<output_index>"    形式的表示还是很长知识的。

注:

这里传给  feed_dict  的变量都是使用 tf.placeholder生成的 tensor 的变量, 这种变量也是整个图所依赖的起始tensor的变量。

-----------------------------------------------------

以下给出 feed_dict 的两个混合写法的 代码:

import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)

def dense(x, size, scope):
    return tf.contrib.layers.fully_connected(x, size, 
                        activation_fn=None, scope=scope)

def dense_relu(x, size, scope):
    with tf.variable_scope(scope):
        h1 = dense(x, size, 'dense')
        return tf.nn.relu(h1, 'relu')


tf.reset_default_graph()
x = tf.placeholder('float32', (None, 784), name='x')
y = tf.placeholder('float32', (None, 10), name='y')
phase = tf.placeholder(tf.bool, name='phase')

h1 = dense_relu(x, 100, 'layer1')
h1 = tf.contrib.layers.batch_norm(h1, 
                            center=True, scale=True, 
                            is_training=phase,
                            scope='bn_1')

h2 = dense_relu(h1, 100, 'layer2')
h2 = tf.contrib.layers.batch_norm(h2, 
                            center=True, scale=True, 
                            is_training=phase,
                            scope='bn_2')

logits = dense(h2, 10, scope='logits')

with tf.name_scope('accuracy'):
    accuracy = tf.reduce_mean(tf.cast(
            tf.equal(tf.argmax(y, 1), tf.argmax(logits, 1)), 
            'float32'))

with tf.name_scope('loss'):
    loss = tf.reduce_mean(
        tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=y))


def train():
    update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
    with tf.control_dependencies(update_ops):
        train_step = tf.train.GradientDescentOptimizer(0.01).minimize(loss)


    gpu_options = tf.GPUOptions( per_process_gpu_memory_fraction=0.5, allow_growth = True )
    sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
    #sess = tf.Session()
    sess.run(tf.global_variables_initializer())
    
    history = []
    iterep = 500
    for i in range(iterep * 30):
        x_train, y_train = mnist.train.next_batch(100)
        sess.run(train_step,
                 feed_dict={x: x_train, 
                            'y:0': y_train, 
                            phase: 1})
        if (i + 1) %  iterep == 0:
            epoch = (i + 1)/iterep
            tr = sess.run([loss, accuracy], 
                          feed_dict={'x:0': mnist.train.images,
                                     y: mnist.train.labels,
                                     phase: 1})
            t = sess.run([loss, accuracy], 
                         feed_dict={x: mnist.test.images,
                                    y: mnist.test.labels,
                                    'phase:0': 0})
            history += [[epoch] + tr + t]
            print( history[-1] )
    return history


train()
原文地址:https://www.cnblogs.com/devilmaycry812839668/p/13466307.html