My laptop:
Ubuntu 16.04 with GPX 1070 (System76, Oryx Pro)
[http://support.system76.com/articles/cuda/]
From [https://developer.nvidia.com/cuda-downloads]
$ wget https://developer.nvidia.com/compute/cuda/8.0/Prod2/local_installers/cuda-repo-ubuntu1604-8-0-local-ga2_8.0.61-1_amd64-deb
$ sudo dpkg -i cuda-repo-ubuntu1604-8-0-local-ga2_8.0.61-1_amd64-deb
[https://github.com/NVIDIA/nvidia-docker/wiki/Installation]
$ sudo apt update
$ sudo apt upgrade
$ sudo apt install cuda
Verify
$ nvidia-smi
Wed Mar 22 07:54:12 2017
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 375.39 Driver Version: 375.39 |
|-------------------------------+----------------------+----------------------+
| 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 1070 Off | 0000:01:00.0 On | N/A |
| N/A 44C P0 30W / N/A | 1814MiB / 8080MiB | 3% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 1370 G /usr/lib/xorg/Xorg 1153MiB |
| 0 3906 G ...s-passed-by-fd --v8-snapshot-passed-by-fd 117MiB |
| 0 17228 G compiz 540MiB |
+-----------------------------------------------------------------------------+
The included source code references a hardcoded prior nvidia driver. You must update this. For the 375 driver version, update by
find /usr/local/cuda-8.0/samples -type f -exec sed -i 's/nvidia-3../nvidia-375/g' {} +
The libraries that are linked by the examples are in /usr/lib/nvidia-375/
Copy all the files to a local folder and run 'make'
(Do NOT install docker.io)
https://docs.docker.com/engine/installation/linux/ubuntu/
Install packages to allow apt to use a repository over HTTPS:
$ sudo apt-get install \
apt-transport-https \
ca-certificates \
curl \
software-properties-common
(Nothing new installed)
$ curl -fsSL https://apt.dockerproject.org/gpg | sudo apt-key add -
$ sudo apt-key fingerprint 58118E89F3A912897C070ADBF76221572C52609D
$ sudo add-apt-repository \
"deb https://apt.dockerproject.org/repo/ \
ubuntu-$(lsb_release -cs) \
main"
New repository is added to
$ more /etc/apt/sources.list
Update sources and install docker-engine
$ sudo apt-get update
$ sudo apt-get -y install docker-engine
Verify
docker -v
Docker version 17.03.0-ce, build 60ccb22
It is still called xxx-ce. But nvidia-docker works with it
Make sure you have the right permission to run the docker binaries. Adding your user account to the group 'docker'
$ sudo usermod -aG docker {user}
Verifiy with 'groups {user}'
Test if you can run the following command without any issues. (I need a logout and login. There must be a better way for that)
$ docker stats
$ wget -P /tmp https://github.com/NVIDIA/nvidia-docker/releases/download/v1.0.1/nvidia-docker_1.0.1-1_amd64.deb
$ sudo dpkg -i /tmp/nvidia-docker*.deb && rm /tmp/nvidia-docker*.deb
verify with running a simple docker test
$ nvidia-docker run --rm nvidia/cuda nvidia-smi
git clone https://github.com/fchollet/keras.git
cd keras\docker
Create an interactive python shell
make ipython