Mandarin

English | 中文

Table of Contents

• 安裝 / 移除 CUDA and cudnn
  • 移除
  • 安裝 CUDA
  • 安裝 cudnn
  • 檢查
• 安裝 TensorRT

---

安裝 / 移除 CUDA 和 cudnn

移除

(我原本的版本是10.1，所以在這邊我是移除10.1的資料夾)

sudo apt-get remove cuda-10.1 
sudo apt autoremove

然後去 /etc/apt/sources.list.d把裡面的cuda相關的檔案刪掉

sudo rm cuda.list 

安裝 CUDA

1. 先從這邊下載deb檔案下來 https://developer.nvidia.com/cuda-10.0-download-archive?target_os=Linux&target_arch=x86_64&target_distro=Ubuntu&target_version=1804&target_type=deblocal
2. sudo dpkg -i cuda-repo-ubuntu1804-10-0-local-10.0.130-410.48_1.0-1_amd64.deb
3. sudo apt-key add /var/cuda-repo-10-0-local-10.0.130-410.48/7fa2af80.pub
4. sudo apt-get update
5. sudo apt-get install cuda
6. sudo apt-get install cuda-libraries-dev-10-0

Other installation options are available in the form of meta-packages. For example, to install all the library packages, replace "cuda" with the "cuda-libraries-10-0" meta package. For more information on all the available meta packages click here.

7. sudo apt-get install cuda-libraries-10-0
8. sudo apt-get install cuda-runtime-10-0
9. sudo apt-get install cuda-toolkit-10-0
10. sudo apt-get install cuda-10-0

記得修改zshrc or bashrc檔案裡面的cuda路徑

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-10.0/lib64
export CUDA_INSTALL_DIR=/usr/local/cuda-10.0
export PATH=$PATH:/usr/local/cuda-10.0/bin
export CUDA_HOME=$CUDA_HOME:/usr/local/cuda-10.0
export PATH=/usr/local/cuda-10.0/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

安裝 cudnn

Download from : https://developer.nvidia.com/rdp/cudnn-download (最好下載tar檔)

複製檔案過去

> sudo cp cuda/include/cudnn.h /usr/local/cuda/include/
> sudo cp cuda/lib64/lib* /usr/local/cuda/lib64/

切換到/usr/local/cuda/lib64/文件夾下

cd /usr/local/cuda/lib64/

建立軟鍊結（需要把版本號換成自己的版本號）

sudo chmod +r libcudnn.so.7.6.5
sudo ln -sf libcudnn.so.7.3.1 libcudnn.so.7
sudo ln -sf libcudnn.so.7 libcudnn.so
sudo ldconfig

檢查

nvidia-smi

nvcc -V

---

安裝 TensorRT

目前最新版本是第七版，如果要下載TensorRT7可以從這邊. (需要先登入帳號)

我的系統是Ubunty 18.04, cuDNN version 7.6.5 and CUDA version 10.0. 我建議是安裝tar包。

我是選擇tar包，如果你的系統環境跟我一樣，可以直接從這裡下載 TensorRT 7.0.0.11 for Ubuntu 18.04 and CUDA 10.0 tar package

解壓縮

tar -zxvf TensorRT-7.0.0.11.Ubuntu-18.04.x86_64-gnu.cuda-10.0.cudnn7.6.tar.gz

安裝步驟請看這邊. 根據我的case, 我是按照這部份.

第一，先進去TensorRT資料夾

cd TensorRT7

安裝 Python TensorRT wheel檔

cd ./python
sudo pip3 install tensorrt-*-cp3x-none-linux_x86_64.whl
cd ..

安裝 Python UFF wheel file. (如果你要用TensorRT跑Tensorflow)

cd ./uff
sudo pip3 install uff-0.6.5-py2.py3-none-any.whl
which convert-to-uff
cd ..

安裝 Python graphsurgeon wheel檔

cd ./graphsurgeon
sudo pip3 install graphsurgeon-0.4.1-py2.py3-none-any.whl
cd ..

Export TensorRT 的library

1. 打開你的 .bashrc / .zshrc

   vim ~/.bashrc
   

   or

   vim ~/.zshrc
   

2. 把你的路徑填進去

   LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/(your location)/TensorRT7/lib
   export TRT_RELEASE=/home/(your location)/TensorRT7_cuda100
   

3. Source

   source ~/.bashrc
   

   or

   source ~/.zshrc
   

檢查

在終端器上直接用python3來做初步測試

import tensorrt

應該不會回報任何錯誤。 再來，你也可以從TensorRT包裡的範例做測試C++版~/TensorRT7/samples/ 和 python版本的~/TensorRT7/samples/python