DokuWiki

User Tools

Site Tools

Trace:
cluster:187

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
cluster:187 [2019/12/10 19:01]
hmeij07 		cluster:187 [2020/08/17 12:01] (current)
hmeij07
Line 1: Line 1:
 \\ \\
 **[[cluster:0|Back]]** **[[cluster:0|Back]]**
 +
 +Slurm links:
 +
 +  * https://slurm.schedmd.com/SLUG19/NVIDIA_Containers.pdf
 +  * https://devblogs.nvidia.com/how-to-run-ngc-deep-learning-containers-with-singularity/
 +  * https://devblogs.nvidia.com/automating-downloads-ngc-container-replicator/
 +  * https://devblogs.nvidia.com/docker-compatibility-singularity-hpc/
 +
 +Other useful links.
 +
 +  * https://www.nvidia.com/en-us/gpu-cloud/containers/
 +  * https://docs.nvidia.com/ngc/ngc-user-guide/index.html
 +    *  scheduler wrapper, inside container: NV_GPU=2,3 nvidia-docker run ...
 +    *  (container sees host gpu 2,3 as container gpu 0,1)
 +  * https://ngc.nvidia.com/catalog/containers/
 +  * https://blog.exxactcorp.com/installing-using-docker-nv-docker-centos-7/
 +  * https://github.com/nvidia/nvidia-container-runtime#nvidia_visible_devices 
 +    *   nvidia? or cuda_visible...
  
  
 ==== NGC Docker Containers ==== ==== NGC Docker Containers ====
  
-Trying to understand how to leverage GPU ready applications on the Nvidia NGC web site (Nvidia GPU Cloud). Download docker containers and buidl your own on premise catalog. Can't wrap myself around the problem of how to integrate containers with the Openlava scheduler.+Trying to understand how to leverage GPU ready applications on the Nvidia NGC web site (Nvidia GPU Cloud). Download docker containers and build your own on premise catalog. Run GPU ready software on compute nodes with docker containers. Can't wrap myself around the problem of how to integrate containers with the our scheduler yet.
  
   * https://blog.exxactcorp.com/installing-using-docker-nv-docker-centos-7/   * https://blog.exxactcorp.com/installing-using-docker-nv-docker-centos-7/
Line 11: Line 29:
 <code> <code>
  
-# Assumes CentOS 7 +# get docker on centos 7 
-# Assumes NVIDIA Driver is installed as per requirements ( < 340.29 ) +curl -fsSL https://get.docker.com/ -o get-docker.sh 
-# Install DOCKER +./getdocker.sh 
-sudo curl -fsSL https://get.docker.com/ | sh + 
-# Start DOCKER +# systemctl 
-sudo systemctl start docker +systemctl enable docker 
-# Add dockeruser, usermod change +systemctl start docker 
-sudo adduser dockeruser+ 
 +# dockeruser, usermod change 
 +adduser dockeruser
 usermod -aG docker dockeruser usermod -aG docker dockeruser
-# Install NV-DOCKER + 
-# GET NVIDIA-DOCKER +# get nvidia-docker 
-wget -P /tmp https://github.com/NVIDIA/nvidia-docker/releases/download/v1.0.1/nvidia-docker-1.0.1-1.x86_64.rpm +# wget -P /tmp https://github.com/NVIDIA/nvidia-docker/releases/download/v1.0.1/nvidia-docker-1.0.1-1.x86_64.rpm 
-# INSTALL +wget https://github.com/NVIDIA/nvidia-docker/archive/v2.2.2.tar.gz 
-sudo rpm -i /tmp/nvidia-docker*.rpm + 
-# Start NV-DOCKER Service+# rpm -i /tmp/nvidia-docker*.rpm 
 +# make nvidia-docker 
 + 
 +# systemctl 
 +systemctl enable nvidia-docker
 systemctl start nvidia-docker systemctl start nvidia-docker
  
Line 103: Line 127:
 # mount host directory to container location # mount host directory to container location
  
--v /home/hmeij:/user/local/hmeij+-v $HOME:/tmp/$USER
  
 # pull images # pull images
Line 124: Line 148:
 # custom build images ... # custom build images ...
 # looks complex based on Dockerfile config file commands # looks complex based on Dockerfile config file commands
-# see link+# see link  
 </code> </code>
  
cluster/187.1576004513.txt.gz · Last modified: 2019/12/10 19:01 by hmeij07

Page Tools

Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
CC Attribution-Share Alike 4.0 International Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki