序言

tensorflow中,為了編譯並執行能夠使用 GPU 的 TensorFlow, 需要先安裝 NVIDIA 提供的 Cuda Toolkit和 CUDNN

失敗的安裝過程

首先tensorflow中文社群的安裝提示中提示Cuda Toolkit 7.0和 CUDNN 6.5 V2.這版本.在CUDA官網中查詢release notes:http://docs.nvidia.com/cuda/#axzz4g719e0em,得知Cuda Toolkit主要包含一下內容:
Compiler
    The CUDA-C and CUDA-C++ compiler
Tools
    The following development tools are available in the bin/ directory (except for Nsight Visual Studio Edition (VSE) which is installed as a plug-in to Microsoft Visual Studio)
    IDEs: nsight (Linux, Mac), Nsight VSE (Windows)
    Debuggers: cuda-memcheck, cuda-gdb (Linux, Mac), Nsight VSE (Windows)
    Profilers: nvprof, nvvp, Nsight VSE (Windows)
    Utilities: cuobjdump, nvdisasm, gwiz
 

cudnn主要是NIVAD提供的神經網路GPU加速的庫.

我選擇安裝最新版本,想當然也知道,N卡加速肯定有系統要求的,去官網檢視安裝說明裡面有如下內容:
To use CUDA on your system, you will need the following installed:

    CUDA-capable GPU
    A supported version of Linux with a gcc compiler and toolchain
    NVIDIA CUDA Toolkit (available at http://developer.nvidia.com/cuda-downloads) 

顯示卡需求:TensorFlow 的 GPU 特性只支援 NVidia Compute Capability >= 3.5 的顯示卡.
作業系統需求和GCC版本需求.
最後才是安裝開發TOOLKIT.
我選擇runfile的安裝形式,官網提示直接使用 sudo sh cudaxxxx


報錯:
It appears that an X server is running. Please exit X before installation. If you're sure that X is not running, but are getting this error, please delete any X lock files in /tmp.
官網提示:
Disable the Nouveau drivers.
Reboot into text mode (runlevel 3).
Verify that the Nouveau drivers are not loaded. If the Nouveau drivers are still loaded, consult your distribution's documentation to see if further steps are needed to disable Nouveau.

其中disable需要將模組新增黑名單,然後載入到核心中.期間需要使用到mkinitramfs

報錯:

E: Problem with MergeList /var/lib/apt/lists/ppa.launchpad.net_vincent-c_nevernote_ubuntu_dists_xenial_main_binary-amd64_Packages

解決:sudo rm /var/lib/apt/lists/* -vf

lsmod | grep nouveau

顯示已經沒有該模組,報錯如下

please delete any x lock file in /tmp

刪除/temp 下.X0檔案
 

ERROR: The kernel module failed to load, because it was not signed by a key

   that is trusted by the kernel. Please try installing the driver

   again, and sign the kernel module when prompted to do so.

仔細看下報錯,然後閱讀安裝說明.

檢視/usr/src/ 目錄下沒有kernel source

使用apt-get install linux-source 4.8.0安裝

放棄了安裝預設361版本的驅動，去官網上下載375的適配我設配的版本。安裝後NVIDIA驅動問題，導致桌面迴圈登入（不確定是不是這個原因，我猜的）

現在桌面進不去，首先我想先把NVIDIA的驅動給刪掉，重啟試試。

命令列下中文亂碼（也是醉了，把語言還原成英文算了）

用vi(或nano等文字編輯器)開啟 /etc/default/locale 檔案
將原來的配置內容修改為
LANG=”en_US.UTF-8″
LANGUAGE=”en_US:en”

Use the following command to uninstall a Driver runfile installation:

$ sudo /usr/bin/nvidia-uninstall

reboot 繼續報錯，session 會話下的所有內容都不會被儲存。此時startx 可以進去了

xsession-error 裡面全都是sogoupy的錯誤？

安裝gdm後命令行也無法進入。放棄了~

重灌系統待續。~

重新整理安裝步驟

經過上述過程，我的ubuntu系統已經崩潰掉了。解決方法如下。
1.重灌Ubuntu16.04系統。
2.安裝NVIDIA375驅動，仍然進不去。
3.解除安裝sudo ./NVIDIAxxxxxxx.run –uninstall 後可以進去。
根據別人部落格說明，解決如下
1.lsmod |grep nouveau，檢視到源生的驅動在。
2.禁用自帶的 nouveau nvidia驅動 （important！）
建立一個檔案通過命令 sudo vim /etc/modprobe.d/blacklist-nouveau.conf
並新增如下內容：
blacklist nouveau
options nouveau modeset=0
再更新一下
sudo update-initramfs -u
修改後需要重啟系統。確認下Nouveau是已經被你幹掉，使用命令： lsmod | grep nouveau
3.禁掉lightdm桌面管理器，安裝驅動（secure boot disabled）
sudo /etc/init.d/lightdm stop
sudo ./NVIDIA-Linux-x86_64-375.20.run –no-opengl-files（這個引數不知道什麼用，別人說有用我就加了）
sudo /etc/init.d/lightdm start
4.OK,解決。

PS:其中用了sigh the kernel 選項，因為without it 安裝失敗。

安裝cuda toolkit

sudo sh cuda_xxxx_linux.run
選項如下

Description

This package includes over 100+ CUDA examples that demonstrate
various CUDA programming principles, and efficient CUDA
implementation of algorithms in specific application domains.
The NVIDIA CUDA Samples License Agreement is available in
Do you accept the previously read EULA?
accept/decline/quit: accept

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 367.48?
(y)es/(n)o/(q)uit: n

Install the CUDA 8.0 Toolkit?
(y)es/(n)o/(q)uit: y

Enter Toolkit Location
 [ default is /usr/local/cuda-8.0 ]:

Do you want to install a symbolic link at /usr/local/cuda?
(y)es/(n)o/(q)uit: y

Install the CUDA 8.0 Samples?
(y)es/(n)o/(q)uit: y

Enter CUDA Samples Location
 [ default is /home/c302 ]:

Installing the CUDA Toolkit in /usr/local/cuda-8.0 ...
Installing the CUDA Samples in /home/c302 ...
Copying samples to /home/c302/NVIDIA_CUDA-8.0_Samples now...
Finished copying samples.

===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installed in /usr/local/cuda-8.0
Samples:  Installed in /home/c302

Please make sure that
 -   PATH includes /usr/local/cuda-8.0/bin
 -   LD_LIBRARY_PATH includes /usr/local/cuda-8.0/lib64, or, add /usr/local/cuda-8.0/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-8.0/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-8.0/doc/pdf for detailed information on setting up CUDA.

***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 361.00 is required for CUDA 8.0 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
    sudo <CudaInstaller>.run -silent -driver

Logfile is /tmp/cuda_install_9045.log

新增環境變數

export PATH=/usr/local/cuda-8.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64:$LD_LIBRARY_PATH

新增系統變數修改到系統檔案

sudo vi /etc/profile
reboot(重啟生效)

驗證cuda

[email protected]-dl:~/Downloads$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2016 NVIDIA Corporation
Built on Sun_Sep__4_22:14:01_CDT_2016
Cuda compilation tools, release 8.0, V8.0.44

測試

cd ‘/home/xxxx/NVIDIA_CUDA-8.0_Samples’
make  //這裡需要點時間
cd 0_Simple/matrixMul
./matrixMul

其他問題：
迴圈登入
黑屏
驅動不工作參見如下部落格
http://www.cnblogs.com/matthewli/p/6715553.html?utm_source=tuicool&utm_medium=referral
secure boot情況下簽名核心模組參考如下：
http://www.linuxdiyf.com/linux/20154.html