keras-gpu/tensorflow-gpu运行程序没有调用GPU问题

安装 keras-gpu或者tensorflow-gpu后运行程序，发现没有调用GPU加速。

运行测试程序

import os
os.environ['CUDA_VISIBLE_DEVICES'] = "0"
import tensorflow as tf
# Creates a graph.
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)
# Creates a session with log_device_placement set to True.
sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))
# Runs the op.
print(sess.run(c))

报错：

failed call to cuInit: CUDA_ERROR_UNKNOWN

原因可能在于CUDA没有正确安装，运行CUDA Samples

cd NVIDIA_CUDA-9.2_Samples
make -j
cd ./bin/x86_64/linux/release
sudo ./deviceQuery

报错：

CUDA driver version is insufficient for CUDA runtime version

说明显卡驱动与CUDA驱动不匹配，更新显卡驱动或者降级CUDA，问题解决。