Tensorflow 没有使用 GPU，发现 xla_gpu 不是 gpu

Question

I just started to explore AI and never used Tensorflow, even Linux is new to me.

I have previously installed NVIDIA Driver 430. It comes with CUDA 10.1

Since Tensorflow-gpu 1.14 doesn't support CUDA 10.1, I uninstalled CUDA 10.1 and I downloaded CUDA 10.0

cuda_10.0.130_410.48_linux.run

once installed I ran

nvcc --version nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2018 NVIDIA Corporation Built on Sat_Aug_25_21:08:01_CDT_2018 Cuda compilation tools, release 10.0, V10.0.130 When I tried to use GPU in Jupyter Notebook, the code still doesn't work

import tensorflow as tf
with tf.device('/gpu:0'):
    a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
    b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
    c = tf.matmul(a, b)

with tf.Session() as sess:
    print (sess.run(c))

Error:

---------------------------------------------------------------------------
InvalidArgumentError                      Traceback (most recent call last)
~/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py in _do_call(self, fn, *args)
   1355     try:
-> 1356       return fn(*args)
   1357     except errors.OpError as e:

~/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py in _run_fn(feed_dict, fetch_list, target_list, options, run_metadata)
   1338       # Ensure any changes to the graph are reflected in the runtime.
-> 1339       self._extend_graph()
   1340       return self._call_tf_sessionrun(

~/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py in _extend_graph(self)
   1373     with self._graph._session_run_lock():  # pylint: disable=protected-access
-> 1374       tf_session.ExtendSession(self._session)
   1375 

InvalidArgumentError: Cannot assign a device for operation MatMul: {{node MatMul}}was explicitly assigned to /device:GPU:0 but available devices are [ /job:localhost/replica:0/task:0/device:CPU:0, /job:localhost/replica:0/task:0/device:XLA_CPU:0, /job:localhost/replica:0/task:0/device:XLA_GPU:0 ]. Make sure the device specification refers to a valid device.
     [[MatMul]]

During handling of the above exception, another exception occurred:

InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-19-3a5be606bcc9> in <module>
      6 
      7 with tf.Session() as sess:
----> 8     print (sess.run(c))

~/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py in run(self, fetches, feed_dict, options, run_metadata)
    948     try:
    949       result = self._run(None, fetches, feed_dict, options_ptr,
--> 950                          run_metadata_ptr)
    951       if run_metadata:
    952         proto_data = tf_session.TF_GetBuffer(run_metadata_ptr)

~/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py in _run(self, handle, fetches, feed_dict, options, run_metadata)
   1171     if final_fetches or final_targets or (handle and feed_dict_tensor):
   1172       results = self._do_run(handle, final_targets, final_fetches,
-> 1173                              feed_dict_tensor, options, run_metadata)
   1174     else:
   1175       results = []

~/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py in _do_run(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)
   1348     if handle is None:
   1349       return self._do_call(_run_fn, feeds, fetches, targets, options,
-> 1350                            run_metadata)
   1351     else:
   1352       return self._do_call(_prun_fn, handle, feeds, fetches)

~/anaconda3/lib/python3.7/site-packages/tensorflow/python/client/session.py in _do_call(self, fn, *args)
   1368           pass
   1369       message = error_interpolation.interpolate(message, self._graph)
-> 1370       raise type(e)(node_def, op, message)
   1371 
   1372   def _extend_graph(self):

InvalidArgumentError: Cannot assign a device for operation MatMul: node MatMul (defined at <ipython-input-9-b145a02709f7>:5) was explicitly assigned to /device:GPU:0 but available devices are [ /job:localhost/replica:0/task:0/device:CPU:0, /job:localhost/replica:0/task:0/device:XLA_CPU:0, /job:localhost/replica:0/task:0/device:XLA_GPU:0 ]. Make sure the device specification refers to a valid device.
     [[MatMul]]

Errors may have originated from an input operation.
Input Source operations connected to node MatMul:
 b (defined at <ipython-input-9-b145a02709f7>:4)    
 a (defined at <ipython-input-9-b145a02709f7>:3)

But, if I ran this code from Terminal in Python, it works. I can see the output

[[22. 28.] [49. 64.]]

Answer 1

You need to make sure you have the appropriate CUDA AND CuDNN versions installed.

• You can check your CuDNN version with the advice from this link: How to verify CuDNN installation?
  • or by running cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2 cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2 on a linux machine
• You can check you CUDA version here: xcat.docs
  • nvcc -V
  • or by running nvidia-smi
• And read about xla_gpu s here: tensorflow xla and here: github xla_gpu issue
  • xla was made by tensorflow, and is faster than standard tensorflow.
  • I'm not sure why CUDA without CuDNN calls gpu s xla_gpu s. Nvidia gpus need CUDA and CuDNN to work properly with Tensorflow, so it looks like tensorflow is trying to use its own library to compute on the GPU. But, I'm not really sure.