install tensorflow on fedora 28.md

add python site

export PYTHONPATH=${PYTHONPATH}:/usr/lib64/python3.6

install cuda and nvdia driver

dnf config-manager --add-repo=http://negativo17.org/repos/fedora-nvidia.repo
# yum-config-manager --add-repo=http://negativo17.org/repos/epel-nvidia.repo
dnf install cuda-devel cuda-cudnn-devel
dnf install nvidia-settings kernel-devel dkms-nvidia vulkan nvidia-driver-libs nvidia-driver-cuda

install tensorflow gpu

conda install -c anaconda tensorflow-gpu

test

import tensorflow as tf
sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))

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)
print(sess.run(c))

check if gpu is there

dnf install nvidia-driver-cuda installs CUDA version 9.2 from the the negativo17 repo. Right?
And tensorflow requires 9.0?