Steps to install and set up Fedora Workstation on my desktop machine.
- Notes from the previous Fedora 20 Installation
#
# /etc/fstab
# Created by anaconda on Sat Jun 16 23:18:44 2018
#
# Accessible filesystems, by reference, are maintained under '/dev/disk'
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more info
#
/dev/mapper/SSD--M2-root / ext4 defaults,noatime 1 1
UUID=ee59ca64-544b-441b-82c6-58b88c43172a /boot ext4 defaults,noatime 1 2
UUID=4AFB-A0F0 /boot/efi vfat umask=0077,shortname=winnt 0 2
/dev/mapper/HDD--2-var /var ext4 defaults,noatime 1 2
/dev/mapper/HDD--2-swap swap swap defaults 0 0
/dev/mapper/luks-7affe2a0-aff3-48f0-b0e8-0c166ede95d3 /home ext4 defaults,noatime,x-systemd.device-timeout=0 1 2
# space --------------------------------------------------------------
/dev/mapper/HDD--1-backup /space/backup ext4 defaults,noatime 1 2
# windows 10 ---------------------------------------------------------
UUID=C880F74780F73A90 /win10/recovery ntfs default,ro 0 0
UUID=DA062A4E062A2BC7 /win10/os ntfs default,ro 0 0
UUID=53EFC79A0A554590 /win10/data ntfs default,rw,uid=1000,gid=1000 0 0
UUID=30DDD99F63BFA995 /win10/temp ntfs default,rw,uid=1000,gid=1000 0 0
# --------------------------------------------------------------------
# eof
#
- Create a user account for family member, match the accounts on my other machines I have. No central user database as I don't want to have to run a server 24/7 to host LDAP, Kerberos, NIS, etc. Add my user to sudoers list.
- Restore user home directories, be careful to selectively skip the hidden ".something" files. In the past I had problems due to these outdated config files when upgrading to newer version of Gnome desktop, Gnome Evolution an other software packagges
- Restore all additional user filesystems from the backup files -- Downloads, Documents, Multimedia, etc.
- Install NVIDIA driver for Linux, choose Beta or Stable version
- Download CUDA Toolkit 9.2
- Choose local installer and download the toolkit and all available patches
- Remove previos installations, if another installation method had been used
- Run
sudo sh cuda_9.2.148_396.37_linux.run
and do not agree to install a video driver, only the toolkit - Update LD_LIBRARY_PATH to point to the CUDA lib folder
- Run nvidia-smi to inspect the GPU, then build the examples and run some
- Install cuDNN 7.1.4 -- NVIDIA Deep Learning SDK, requires free account with NVIDIA Developer Program
- Install NCCL 2.2.13 -- NVIDIA Collective Communications Library
- Install Anaconda to use Conda environments and channels for these tasks
- For Intel processors, use this Intel channel to Optimize your CPU for Deep Learning
> conda create -n intel -c intel intelpython3_full
> conda activate intel
> conda install tensorflow-mkl
> pip install intel-tensorflow
> conda install keras -c intel
> conda install jupyter -c intel
- Build TensorFlow from source
- Create conda environment with GCC 7.1 and Python 3.6
- Download and install bazel
- Kick off the build which is quite long
- Install PyTorch and Keras
- How to install Tensorflow GPU with CUDA 9.2 for python on Ubuntu
- Up and Running with Ubuntu, Nvidia, Cuda, CuDNN, TensorFlow, and PyTorch
If using Chrome browser start with GNOME Shell integration, for Debian, Fedora, Gentoo and Ubuntu you can install package named "chrome-gnome-shell". This extension provides integration with GNOME Shell and the corresponding extensions repository https://extensions.gnome.org
- Hide top bar
- Sound Input & Output Device Chooser
- Freon
- Recent Items
- No Title Bar
- Gnote/Tomboy Integration
- Bing Wallpaper Changer
- Media Player Indicator
- Adjust Brightness Icon
To list all installed extensions and to manage them, go to installed extensions.
> dnf install gnome-tweak-tool
> dnf search gtk | grep theme
> dnf search [shell-theme | icon-theme | cursor-theme]