安装CUDA和驱动

Novu

步骤

• SuperMicro的机器必须, 在linux项的quiet splash后面加上nomodeset.

• 清理旧的driver

sudo apt-get purge nvidia-*
sudo apt-get autoremove
sudo reboot

• 关闭lightdm: sudo service lightdm stop
• 安装Nvidia driver. 加入ppa直接安装nvidia-384驱动. (必须单独安装Nvidia driver, 不要用CUDA里的显卡驱动!!!)

sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-get update
sudo apt-get install nvidia-384
sudo reboot

检查一下nvidia-smi
* 下载CUDA 8. 千万记住不要选择driver和opengl!

wget https://developer.nvidia.com/compute/cuda/8.0/Prod2/local_installers/cuda_8.0.61_375.26_linux-run
sudo sh cuda_8.0.61_375.26_linux.run --override --no-opengl-lib

安装cuDNN

• 用Caffe的脚本安装cuDNN. ```

CUDNN_TAR_FILE="cudnn-8.0-linux-x64-v6.0.tgz"
sudo apt-get upgrade -y
sudo apt-get install -y opencl-headers build-essential protobuf-compiler \
    libprotoc-dev libboost-all-dev libleveldb-dev hdf5-tools libhdf5-serial-dev \
    libopencv-core-dev  libopencv-highgui-dev libsnappy-dev \
    libatlas-base-dev cmake libstdc++6-4.8-dbg libgoogle-glog0v5 libgoogle-glog-dev \
    libgflags-dev liblmdb-dev git python-pip gfortran libopencv-dev
sudo apt-get clean

• (optional) disable secure boot.

sudo apt install mokutil
sudo mokutil --disable-validation

安装nvidia-docker出问题: no UVM module

• 老问题了: Error: Could not load UVM kernel module. Is nvidia-modprobe installed?
• 奇怪, sudo apt-get install nvidia-384应该会自动安装modprobe啊. 没事, sudo apt-get install nvidia-modprobe.
• 然后sudo nvidia-modprobe -u -c=0. 额. nvidia-docker-plugin不work, 看了一下service nvidia-docker status已经error了. 因为刚刚nvidia-modprobe之后还要service nvidia-docker restart. 再看一下status. 果然好了.

大功告成!

参考

• 额
• 最主要参考
• 额额

什么是linux-hwe

• update之后, 安装CUDA就报错 kernel path not found.

p1@p1-M32CD:~$ ll /usr/src
total 149892
drwxr-xr-x  5 root root      4096 Oct 27 11:52 ./
drwxr-xr-x 11 root root      4096 Aug  1 04:24 ../
drwxr-xr-x 27 root root      4096 Oct 26 10:41 linux-headers-4.10.0-37/
drwxr-xr-x  7 root root      4096 Oct 26 10:41 linux-headers-4.10.0-37-generic/
drwxr-xr-x 30 root root      4096 Oct 27 11:20 linux-hwe-4.10.0/
-rw-r--r--  1 root root   9541226 Oct  9 23:59 linux-hwe_4.10.0-37.41~16.04.1.diff.gz
-rw-r--r--  1 root root      5445 Oct  9 23:59 linux-hwe_4.10.0-37.41~16.04.1.dsc
-rw-r--r--  1 root root 143917000 Jun 29 23:08 linux-hwe_4.10.0.orig.tar.gz