Asus主機 i7 Ubuntu16.04 + GTX 1080 8G install driver
Install CUDA10.0 Reference
Step1. 先update and upgrade 一下
$ sudo apt-get update
$ sudo apt-get upgrade
- 選擇 Linux -> x86_64 -> Ubuntu -> 16.04 ->deb(network選比較小的覺得速度較快哈 )
$ `sudo dpkg -i cuda-repo-ubuntu1604_10.0.130-1_amd64.deb`
$ `sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub`
When installing using network repo on Ubuntu 17.10:
$ sudo apt-key adv --fetch-keys https:/...
When installing using network repo on Ubuntu 16.04:
$ sudo apt-key adv --fetch-keys http:/...
Ubuntu 17.10 跟 16.04 路徑只差 https 跟 http！！！
$ `sudo apt-get update`
$ `sudo apt-get install cuda`
Step3. follow 3.7.2 Package_Upgrades
$ sudo apt-get install cuda-drivers
Step4. follow 7.1.1 Environment Setup
$ 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}}
Step5. follow 7.3.1 install Third-party Libraries
(不確定需不需要，都裝＠＠)
sudo apt-get install g++ freeglut3-dev build-essential libx11-dev \
libxmu-dev libxi-dev libglu1-mesa libglu1-mesa-dev
Step6. follow 2.1 Verify you have a CUDA-Capable GPU
lspci | grep -i nvidia
follow 2.4
sudo apt-get install linux-headers-$(uname -r)
Step7. 檢查一些相關...
$ lspci
$ dmesg
$ lsmod
Step8. follow 4.3.5 Ubuntu Disabling Nouveau
created the file:
$ sudo vim /etc/modprobe.d/blacklist-nouveau.conf
(加入以下)
blacklist nouveau
options nouveau modeset=0
$ sudo update-initramfs -u
$ sudo vim /etc/profile
(加入以下)
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}}
Step10. 重開機吧～～～～～～
Step11. (如果開機還能用就繼續 step 11. 執行 Test samples)
follow 7.2.2 install Writeable Samples,會在home下產生
/NVIDIA_CUDA-10.0_Samples
$ `cuda-install-samples-10.0.sh .`
Step12. follow 7.2.3.1 Verify the Driver Version
$ cat /proc/driver/nvidia/version
$ nvcc -V
$ cd NVIDIA_CUDA-10.0_Samples
$ cat /proc/cpuinfo ->可以看到電腦幾核心
$ make -j or (make -j8 電腦8核
Step13. follow 7.2.3.3 Running the Binaries
$ cd ~/NVIDIA_CUDA-10.0_Samples/bin/x86_64/linux/release
$ ./deviceQuery
$ ./bandwidthTest
Install cuDNN7.4.1 Reference
dpkg tips
dpkg -i package.deb #安装包
dpkg -r package #删除包
dpkg -P package #删除包（包括配置文件）
dpkg -L package #列出与该包关联的文件
dpkg -l package #显示该包的版本
dpkg --unpack package.deb #解开deb包的内容
dpkg -S keyword #搜索所属的包内容
dpkg -l #列出当前已安装的包
dpkg -c package.deb #列出deb包的内容
dpkg --configure package #配置包
要註冊才能下載！
follow 2.3. Installing cuDNN on Linux
解壓縮資料夾
$ tar -xzvf cudnn-10.0-linux-x64-v7.4.1.5.tgz
Step2. 去到解壓縮的資料夾裡
解壓縮的資料夾/cuda 裡面還有兩個資料夾
/cuda/lib64/
/cuda/include/
複製裏面資料到 local/cuda 裏面
$ cd ../cuda/lib64
$ sudo cp ./* /usr/local/cuda/lib64/
$ cd ../cuda/include
$ sudo cp cudnn.h /usr/local/cuda/include/
更新他們的軟連結
$ cd /usr/local/cuda/lib64/
$ sudo rm -rf libcudnn.so libcudnn.so.7
$ sudo ln -s libcudnn.so.7.4.1 libcudnn.so.7
$ sudo ln -s libcudnn.so.7 libcudnn.so
Step3. 檢查cudnn 安裝成功與否
安裝：
$ sudo dpkg -i libcudnn7-doc_7.4.1.5-1+cuda10.0_amd64.deb
查安裝路徑：
$ sudo dpkg -c libcudnn7-doc_7.4.1.5-1+cuda10.0_amd64.deb
查到 /mnistCUDNN 的完整路徑在
/usr/src/cudnn_samples_v7/mnistCUDNN
$ cd /usr/src/cudnn_samples_v7/mnistCUDNN
follow 2.4. Verifying
-------------------------------------------
因為這路徑下是unwritable path.所以複製到可讀寫的路徑
$ `cp -r /usr/src/cudnn_samples_v7/ .`
$ `cd ./cudnn_samples_v7/mnistCUDNN`
編譯：
$ make clean & make -j (or make -j8
執行：
$ ./mnistCUDNN
**如果直接在/usr/src/cudnn_samples_v/ 編譯&執行有error發生**
有可能會跑出下列錯誤
./mnistCUDNN: error while loading shared libraries: libcudart.so.8.0:
cannot open shared object file: No such file or directory
$ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64
看一下安裝之後
$ nvidia-smi
Wed Dec 5 14:36:14 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.79 Driver Version: 410.79 CUDA Version: 10.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 1080 Off | 00000000:01:00.0 On | N/A |
| 0% 36C P8 9W / 198W | 378MiB / 8116MiB | 1% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 1124 G /usr/lib/xorg/Xorg 218MiB |
| 0 2044 G compiz 155MiB |
| 0 19569 G /usr/lib/firefox/firefox 2MiB |
+-----------------------------------------------------------------------------+
$ ./bandwidthTest
留言
張貼留言