Skip to content

advdockerguide.md

Latest commit

 

History

History
executable file
·
40 lines (30 loc) · 1.33 KB

advdockerguide.md

File metadata and controls

executable file
·
40 lines (30 loc) · 1.33 KB

Home | DevBox | Athena | Supercomputing@SUTD

Advanced Guide to using nvidia-docker

How to scp files from DGX to local

$ scp username@server:/home/username/directory/you/want /local/directory/you/want/filename

How to scp files from local to DGX

$ scp /local/directory/you/want/filename username@server:/home/username/directory/you/want

Note: for folders, use -r handle.

To save your docker image

username@server:~$ nvidia-docker commit *DOCKERNAMETOCOMMIT* *IMAGENAMETOSAVEAS*

To transfer from host to docker container

username@server:~$ nvidia-docker cp *./nameoffile* *dockername/number*:/*dockerdirectory*
username@server:~$ nvidia-docker cp ./README.md my_pyconda3:/
username@server:~$ nvidia-docker cp ./README.md e74d0:/

the first 5 characters of the container ID is sufficient to identify the container.

To transfer from docker container to host

username@server:~$ nvidia-docker cp *dockername/number*:/*dockerdirectorywithfile* *./*
username@server:~$ nvidia-docker cp my_pyconda3:/README.md ./
username@server:~$ nvidia-docker cp e74d0:/README.md ./

Installation of pycuda

Install Anaconda3, install pip and upgrade it, then pip install pycuda.