TensorflowでGPUが使えてるか確認する

$ lspci | grep -i nvidia
01:00.0 VGA compatible controller: NVIDIA Corporation GM204 [GeForce GTX 970] (rev a1)
01:00.1 Audio device: NVIDIA Corporation GM204 High Definition Audio Controller (rev a1)

$ cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module  418.39  Sat Feb  9 19:19:37 CST 2019
GCC version:  gcc version 7.3.0 (Ubuntu 7.3.0-27ubuntu1~18.04)

$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Fri_Feb__8_19:08:17_PST_2019
Cuda compilation tools, release 10.1, V10.1.105

pip list | grep pycuda
pycuda               2018.1.1

$ apt list | grep tensor
nv-tensorrt-repo-ubuntu1804-cuda10.0-trt5.0.2.6-ga-20181009/now 1-1 amd64 [インストール済み、ローカル]
r-cran-tensor/bionic,bionic 1.5-2 all
xtensor-dev/bionic,bionic 0.10.11-1 all
xtensor-doc/bionic,bionic 0.10.11-1 all
xtensor-python-dev/bionic,bionic 0.12.4-1 all
xtensor-python-doc/bionic,bionic 0.12.4-1 all

$ nvidia-smi
Sat Mar  9 06:45:49 2019
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 418.39       Driver Version: 418.39       CUDA Version: 10.1     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 970     On   | 00000000:01:00.0  On |                  N/A |
|  0%   43C    P0    45W / 151W |    491MiB /  4039MiB |      1%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      1832      G   /usr/lib/xorg/Xorg                           260MiB |
|    0      2002      G   /usr/bin/gnome-shell                         133MiB |
|    0      3022      G   ...quest-channel-token=4667018627475572094    93MiB |
+-----------------------------------------------------------------------------+

from tensorflow.python.client import device_lib
device_lib.list_local_devices()

2019-03-09 06:30:33.408847: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-03-09 06:30:33.439051: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 4008000000 Hz
2019-03-09 06:30:33.439560: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x55bb5a801000 executing computations on platform Host. Devices:
2019-03-09 06:30:33.439575: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>

$ pip uninstall tensorflow
$ pip install --upgrade pip setuptools wheel
$ pip install -I tensorflow-gpu

$ python
Python 3.6.0 (default, Mar  7 2019, 22:28:33)
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
Traceback (most recent call last):
...
ImportError: libcublas.so.10.0: cannot open shared object file: No such file or directory

$ sudo apt install cuda-libraries-10-0
$ sudo apt purge nvidia-driver-418
$ sudo apt-get install --no-install-recommends nvidia-driver-410
$ reboot

$ python gpu_check.py
2019-03-09 09:41:14.378974: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-03-09 09:41:14.516241: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:998] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-03-09 09:41:14.516758: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x5618aba42e90 executing computations on platform CUDA. Devices:
2019-03-09 09:41:14.516777: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): GeForce GTX 970, Compute Capability 5.2
2019-03-09 09:41:14.536402: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 4008000000 Hz
2019-03-09 09:41:14.536878: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x5618aba74050 executing computations on platform Host. Devices:
2019-03-09 09:41:14.536895: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
2019-03-09 09:41:14.537021: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1433] Found device 0 with properties:
name: GeForce GTX 970 major: 5 minor: 2 memoryClockRate(GHz): 1.1775
pciBusID: 0000:01:00.0
totalMemory: 3.94GiB freeMemory: 3.46GiB
2019-03-09 09:41:14.537033: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1512] Adding visible gpu devices: 0
2019-03-09 09:41:14.538418: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-03-09 09:41:14.538432: I tensorflow/core/common_runtime/gpu/gpu_device.cc:990]      0
2019-03-09 09:41:14.538437: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1003] 0:   N
2019-03-09 09:41:14.538532: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/device:GPU:0 with 3244 MB memory) -> physical GPU (device: 0, name: GeForce GTX 970, pci bus id: 0000:01:00.0, compute capability: 5.2)