安裝的前提是你已經安裝好顯卡驅動，如果沒有安裝好，丟個傳送門關於ubuntu19.04下安裝獨立顯卡驅動
安裝Anaconda4.7.10
先去anaconda官網下載，這裏選擇則64-Bit (x86) Installer
然後在下載目錄打開終端，輸入
bash Anaconda3-2019.07-Linux-x86_64.sh
之後會出現協議，這裏不用管，一直Enter到結束，之後出現
Do you approve the license terms? [yes|no]
輸入yes即可，再之後會出現
Anaconda3 will now be installed into this location:
/home/meteorrain/anaconda3
- Press ENTER to confirm the location
- Press CTRL-C to abort the installation
- Or specify a different location below
[/home/meteorrain/anaconda3] >>>
就是選擇安裝的位置，看個人喜好，我是比較懶，直接默認安裝了，然後.....就裝完了(」゜ロ゜)」
並沒有出現是否添加到path的選項....但是我後來查看確實在路徑中，可能是默認添加了= =，或者說你選擇默認安裝，就會自動添加路徑？....不管了，進行下一步，更新系統文件
source ~/.bashrc
然後輸入如下命令
conda --version
顯示
conda 4.7.10
則表示安裝成功了～～
安裝cuda10.1
不需要g++降級，去cuda官網直接下載相應的版本即可，由於暫時沒有19.04的，所以我下載的18.10，其實也沒多大影響，最近幾代ubuntu的兼容性很好
然後就是補庫（我沒有測試這些庫是否有用，但是爲了省事，就都裝上了～）
sudo apt-get install freeglut3-dev libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev
然後再輸入
sudo chmod 777 cuda_10.1.168_418.67_linux.run
sudo cuda_10.1.168_418.67_linux.run
安裝程序第一個輸入accept，第二個直接選擇install，然後等待安裝完成即可～再編輯一下這個文件
sudo vi ~/.bashrc
在文件末尾加上這幾行
export CUDA_HOME=/usr/local/cuda
export PATH=$PATH:$CUDA_HOME/bin
export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
輸入:wq保存並退出，然後輸入如下命令進行更新
source ~/.bashrc
以下命令是用來測試cuda是否配置成功的
輸入
cat /proc/driver/nvidia/version
輸出
NVRM version: NVIDIA UNIX x86_64 Kernel Module 418.67 Sat Apr 6 03:07:24 CDT 2019
GCC version: gcc version 8.3.0 (Ubuntu 8.3.0-6ubuntu1)
輸入
nvcc -V
輸出
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Wed_Apr_24_19:10:27_PDT_2019
Cuda compilation tools, release 10.1, V10.1.168
安裝cudnn7.6.2
cudnn的版本一定要根據cuda的版本進行選擇，官網傳送門，下載cudnn還必須註冊賬號，以及填寫一些信息，這些就隨意啦～
網上搜索的安裝cudnn感覺好麻煩，這裏就是下載的問題了，有人建議下載cuDNN Library for Linux，然後安裝過程大概有五條命令的樣子（這裏就不貼出來了），我選擇的是下載deb安裝包，這裏有兩種安裝包， cuDNN Runtime Library for Ubuntu18.04 (Deb)以及cuDNN Developer Library for Ubuntu18.04 (Deb)（由於沒有針對19.04版本的所以下載的18.04版本），請下載Runtime的，因爲Runtime是基本運行庫，Developer是依賴於Runtime針對有興趣開發cuDNN的人的庫，所以我們只需要下載Runtime就可以了，下載完成後輸入命令
sudo dpkg -i libcudnn7_7.6.2.24-1+cuda10.1_amd64.deb
然後等待安裝完成就可以啦～
這裏補充一下，請不要使用如下命令驗證是否安裝成功
cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
因爲安裝方式不同，驗證方法也是不一樣的，因爲使用deb文件安裝，根本在/usr/local/cuda/include目錄下根本沒有cudnn.h這個文件，正確的驗證方式在官網上給的清清楚楚， 傳送門，當然，這裏要先安裝一下測試用例
sudo dpkg -i libcudnn7-doc_7.6.2.24-1+cuda10.1_amd64.deb
我把官網的驗證操作搬運過來
Copy the cuDNN sample to a writable path.（這一步是必須的，因爲如果你直接cd到cudnn_samples_v7裏執行make操作會沒有權限）
$cp -r /usr/src/cudnn_samples_v7/ $HOME
Go to the writable path.
$ cd $HOME/cudnn_samples_v7/mnistCUDNN
編譯樣例mnistCUDNN
$make clean && make
運行樣例mnistCUDNN
$ ./mnistCUDNN
如果cuDNN正確的安裝，你將會看到和如下相似的結果：
cudnnGetVersion() : 7602 , CUDNN_VERSION from cudnn.h : 7602 (7.6.2)
Host compiler version : GCC 8.3.0
There are 1 CUDA capable devices on your machine :
device 0 : sms 68 Capabilities 7.5, SmClock 1545.0 Mhz, MemSize (Mb) 10986, MemClock 7000.0 Mhz, Ecc=0, boardGroupID=0
Using device 0
Testing single precision
Loading image data/one_28x28.pgm
Performing forward propagation ...
Testing cudnnGetConvolutionForwardAlgorithm ...
Fastest algorithm is Algo 0
Testing cudnnFindConvolutionForwardAlgorithm ...
^^^^ CUDNN_STATUS_SUCCESS for Algo 0: 0.016480 time requiring 0 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 2: 0.030816 time requiring 57600 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 1: 0.042144 time requiring 3464 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 5: 0.047840 time requiring 203008 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 7: 0.057280 time requiring 2057744 memory
Resulting weights from Softmax:
0.0000000 0.9999399 0.0000000 0.0000000 0.0000561 0.0000000 0.0000012 0.0000017 0.0000010 0.0000000
Loading image data/three_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000000 0.0000000 0.9999288 0.0000000 0.0000711 0.0000000 0.0000000 0.0000000 0.0000000
Loading image data/five_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000008 0.0000000 0.0000002 0.0000000 0.9999820 0.0000154 0.0000000 0.0000012 0.0000006
Result of classification: 1 3 5
Test passed!
Testing half precision (math in single precision)
Loading image data/one_28x28.pgm
Performing forward propagation ...
Testing cudnnGetConvolutionForwardAlgorithm ...
Fastest algorithm is Algo 0
Testing cudnnFindConvolutionForwardAlgorithm ...
^^^^ CUDNN_STATUS_SUCCESS for Algo 0: 0.014784 time requiring 0 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 2: 0.014976 time requiring 28800 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 1: 0.034816 time requiring 3464 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 7: 0.042080 time requiring 2057744 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 5: 0.044000 time requiring 203008 memory
Resulting weights from Softmax:
0.0000001 1.0000000 0.0000001 0.0000000 0.0000563 0.0000001 0.0000012 0.0000017 0.0000010 0.0000001
Loading image data/three_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000000 0.0000000 1.0000000 0.0000000 0.0000714 0.0000000 0.0000000 0.0000000 0.0000000
Loading image data/five_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000008 0.0000000 0.0000002 0.0000000 1.0000000 0.0000154 0.0000000 0.0000012 0.0000006
Result of classification: 1 3 5
Test passed!
安裝Tensorflow2.0
emmmm，裝這個是最簡單的，如果上面裝的都和我一樣的話，下面的也請跟我裝的一樣，alpha0版本我測試過似乎有問題= =
pip3 install tensorflow-gpu==2.0.0-beta1
進入python3環境，然後輸入
import tensorflow as tf
這裏有可能會出現這個警告
/home/meteorrain/.local/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:516: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint8 = np.dtype([("qint8", np.int8, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:517: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint8 = np.dtype([("quint8", np.uint8, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:518: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint16 = np.dtype([("qint16", np.int16, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:519: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint16 = np.dtype([("quint16", np.uint16, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:520: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint32 = np.dtype([("qint32", np.int32, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:525: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
np_resource = np.dtype([("resource", np.ubyte, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:541: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint8 = np.dtype([("qint8", np.int8, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:542: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint8 = np.dtype([("quint8", np.uint8, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:543: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint16 = np.dtype([("qint16", np.int16, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:544: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint16 = np.dtype([("quint16", np.uint16, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:545: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint32 = np.dtype([("qint32", np.int32, 1)])
/home/meteorrain/.local/lib/python3.7/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:550: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
np_resource = np.dtype([("resource", np.ubyte, 1)])
可以無視掉，因爲官網的實例中也有，而且這個只是warning不是error，只是告訴你，這樣寫不太合適，究其原因是numpy版本的問題，如果想不顯示，就將numpy版本降級，網上有方法不再贅述，這裏我選擇放着不管（官方都不怕，咱們怕啥(๑•́ ₃•̀๑)）
然後輸入
tf.__version__
如果輸出
'2.0.0-beta1'
就表示，安裝成功了～
安裝Pytorch
首先在pytorch官網選擇自己的環境，python的版本，cuda的版本，以及安裝的方式啥的，然後官網會自動生成一個安裝命令，將其複製到終端執行即可，不過我安裝的時候沒有cuda10.1的選項，於是乎，我自己手動修改了其中的版本號，輸入瞭如下命令：
conda install pytorch torchvision cudatoolkit=10.1 -c pytorch
當然也是沒有問題的，最後測試以下是否安裝成功，在python3環境下輸入
import torch
沒有報錯的話就是安裝成功啦～
這裏附上一個安裝虛擬環境的帖子，這裏是創建的python虛擬環境，
實際上可以直接創建anaconda的虛擬環境，命令如下
conda create --name exp python=3.7.3
exp是自己定義的環境名稱，後面的3.7.3是python的版本，然後會安裝一些庫，安裝完成後，輸入
source activate exp
即可進入虛擬環境進行操作，下面是一些簡單的Anaconda指令
創建環境 conda create -n environment_name python=3
刪除環境 conda remove -n environment_name --all
激活環境 source activate environment_name /conda activate environment_name
退出環境 conda deactivate
安裝包 conda install package
查看已安裝包 conda list
更新包 conda update package
移除包 conda remove package
覺得文章還不錯有幫助的童鞋還請點個贊支持一下～( •̀∀•́ )