一開始接觸到這個配置的時候說實話是十分茫然的，花了很久的時間查資料，好在最後安裝成功了！下面是本人安裝的過程。
首先，在開始之前先看需要裝什麼版本的CUNA，cuDNN，以及tensorflow-gpu的版本。
步驟一：
檢查GPU驅動是否完成：
~$ nvidia-smi
結果如下表示驅動安裝成功：
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 418.40.04 Driver Version: 418.40.04 CUDA Version: 10.1 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla P4 Off | 00000000:03:00.0 Off | 0 |
| N/A 37C P8 6W / 75W | 0MiB / 7611MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
查看服務器的當前個數：
$ lspci|grep -i nvidia
查看結果爲：
03:00.0 3D controller: NVIDIA Corporation GP104GL [Tesla P4] (rev a1)
步驟二：我這裏是已經安裝了GPU驅動，所以安裝的步驟上面沒有介紹。接下來是查看CUDA，cuDNN和tensorflow-gpu的版本對應情況。注意一定要對應，才能安裝成功。
在這裏，我安裝的是第一行的tensorflow-gpu 1.13.1,CUNA10.0,cuDNN7.4.
步驟三：CUDA的安裝
根據如下圖的選擇下載CUNA10.0：
這裏將安裝包和補丁都下載，補丁後面後用到。
下載完成後，用命令進行安裝：
sudo sh cuda_10.0.130_410.48_linux.run
然後會出現以下的安裝內容，如下進行選擇：
Do you accept the previously read EULA?
accept/decline/quit: accept
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 410.48?
(y)es/(n)o/(q)uit: n
Install the CUDA 10.0 Toolkit?
(y)es/(n)o/(q)uit: y
Enter Toolkit Location
[ default is /usr/local/cuda-10.0 ]:
Do you want to install a symbolic link at /usr/local/cuda?
(y)es/(n)o/(q)uit: y
Install the CUDA 10.0 Samples?
(y)es/(n)o/(q)uit: y
Enter CUDA Samples Location
[ default is /home/wanying ]:
接着會出現以下內容：
Installing the CUDA Samples in /home/wanying ...
Copying samples to /home/wanying/NVIDIA_CUDA-10.0_Samples now...
Finished copying samples.
===========
= Summary =
===========
Driver: Not Selected
Toolkit: Installed in /usr/local/cuda-10.0
Samples: Installed in /home/wanying, but missing recommended libraries
Please make sure that
- PATH includes /usr/local/cuda-10.0/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-10.0/lib64, or, add /usr/local/cuda-10.0/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-10.0/bin
Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.0/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 384.00 is required for CUDA 10.0 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_45102.log
接下來將之前下載的補丁也安裝：
sudo sh cuda_10.0.130.1_linux.run
最後配置環境變量，首先打開bashrc文件，將路徑添加在後面：
vim ~/.bashrc
接着對bashrc文件進行編輯，添加下面幾句話，第一行爲註釋，不用添加：
# cuda PATH
export PATH=/usr/local/cuda-10.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64:$LD_LIBRARY_PATH
到這裏，CUDA的安裝完成，接下來，我們進行測試，看是否安裝成功。先進入到測試樣本的目錄下：
$ cd NVIDIA_CUDA-10.0_Samples/1_Utilities/deviceQuery
然後進行測試；
sudo make
./deviceQuery
出現以下結果，測試通過（全部內容很多，只截取了最後的部分）。CUDA安裝完成！
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.1, CUDA Runtime Version = 10.0, NumDevs = 1
Result = PASS
步驟四：cuDNN的安裝
根據對應的版本，下載鏈接爲：https://developer.nvidia.com/rdp/cudnn-archive。我這裏下載的是cuDNN7.4.2，下載cuDNN Library for Linux，
下載完cudnn-10.0-linux-x64-v7.4.2.24之後進行解壓，解壓之後是cuda文件夾，進入cuda/include目錄，在命令行進行如下操作：
sudo cp cudnn.h /usr/local/cuda/include/ #複製頭文件
再進入cuda/lib64目錄下的動態文件進行復制和鏈接（根據自己的版本號修改libcudnn.so）：
sudo cp lib* /usr/local/cuda/lib64/ # 複製動態鏈接庫
cd /usr/local/cuda/lib64/
sudo rm -rf libcudnn.so libcudnn.so.7 # 刪除原有動態文件
sudo ln -s libcudnn.so.7.4.2 libcudnn.so.7 # 生成軟銜接
sudo ln -s libcudnn.so.7 libcudnn.so # 生成軟鏈接
然後對cuDNN是否安裝成功進行測試，先下載三個包，鏈接與下載cuDNN安裝包的鏈接一樣。如下圖：
下載完成後，用命令安裝（我裝的時候沒注意順序出錯了，不確定是不是順序的原因，最好是按照順序來裝）：
$ sudo dpkg -i libcudnn7_7.4.2.24-1+cuda10.0_amd64.deb
$ sudo dpkg -i libcudnn7-dev_7.4.2.24-1+cuda10.0_amd64.deb
$ sudo dpkg -i libcudnn7-doc_7.4.2.24-1+cuda10.0_amd64.deb
安裝完成後，
cp -r /usr/src/cudnn_samples_v7/mnistCUDNN $HOME
cd $HOME/mnistCUDNN
make clean && make
./mnistCUDNN
運行結果如下，表示成功：
Testing half precision (math in single precision)
Loading image data/one_28x28.pgm
Performing forward propagation ...
Testing cudnnGetConvolutionForwardAlgorithm ...
Fastest algorithm is Algo 1
Testing cudnnFindConvolutionForwardAlgorithm ...
^^^^ CUDNN_STATUS_SUCCESS for Algo 0: 0.024576 time requiring 0 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 1: 0.032768 time requiring 3464 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 2: 0.045056 time requiring 28800 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 4: 0.100352 time requiring 207360 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 7: 0.133120 time requiring 2057744 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!
步驟五：tensorflow-gpu的安裝
首先，我們要確保系統中安裝瞭如下Python環境：python3, pip3，以及 virtualenv，在命令行查詢相應的版本
python3 --version
pip3 --version
virtualenv --version
如果缺什麼就按照如下命令進行安裝：（缺什麼安裝什麼）
sudo apt update
sudo apt install python3-dev python3-pip
sudo pip3 install -U virtualenv
接下來安裝python虛擬環境，python的虛擬環境用來隔離系統和相應的安裝包，這非常有利於不同版本之間的隔離，總之好處多多，尤其是不同的項目使用不同的軟件版本時，能避免令人頭痛的版本混亂問題，強烈建議安裝虛擬環境。安裝命令如下：
virtualenv --system-site-packages -p python3 ./venv # 創建python3虛擬環境
source ./venv/bin/activate # 激活虛擬環境
安裝Tensorflow
激活虛擬環境後， 安裝對應版本的tensorflow-gpu， 我的版本是：tensorflow-gpu 1.13.1, 安裝命令如下：
pip install tensorflow-gpu==1.13.1
安裝完成之後，測試是否安裝成功命令如下：
python -c "import tensorflow as tf; tf.enable_eager_execution(); print(tf.reduce_sum(tf.rando
m_normal([1000, 1000])))"
運行結果如下：
2019-07-01 18:25:42.769805: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-07-01 18:25:42.909912: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x182b620 executing computations on platform CUDA. Devices:
2019-07-01 18:25:42.909982: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): Tesla P4, Compute Capability 6.1
2019-07-01 18:25:42.932521: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2399945000 Hz
2019-07-01 18:25:42.937288: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x1eef3d0 executing computations on platform Host. Devices:
2019-07-01 18:25:42.937345: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): <undefined>, <undefined>
2019-07-01 18:25:42.938332: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1433] Found device 0 with properties:
name: Tesla P4 major: 6 minor: 1 memoryClockRate(GHz): 1.1135
pciBusID: 0000:03:00.0
totalMemory: 7.43GiB freeMemory: 7.32GiB
2019-07-01 18:25:42.938377: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1512] Adding visible gpu devices: 0
2019-07-01 18:25:42.940093: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-07-01 18:25:42.940124: I tensorflow/core/common_runtime/gpu/gpu_device.cc:990] 0
2019-07-01 18:25:42.940138: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1003] 0: N
2019-07-01 18:25:42.941039: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 7123 MB memory) -> physical GPU (device: 0, name: Tesla P4, pci bus id: 0000:03:00.0, compute capability: 6.1)
tf.Tensor(-685.63104, shape=(), dtype=float32)
至此，完成tensorflow-gpu版本的配置。 可以愉快玩耍了!!!