[
https://issues.apache.org/jira/browse/SINGA-276?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15769168#comment-15769168
]
wangwei commented on SINGA-276:
-------------------------------
I suggest to use the following naming rule.
{code}
singa:devel|runtime[-OS][-CUDA|OPENCL][-CUDNN]
{code}
devel: development images with all dependent libs' header files installed and
SINGA's source code; runtime: the minimal images which can run SINGA programs.
OS: ubuntu, ubuntu14.04, centos, centos6
CUDA: cuda, cuda8.0, cuda7.0
cudnn: cudnn, cudnn5, cudnn4
OPENCL: opencl, opencl1.2
By default, if the version is not included in the tag, the latest stable
version is used.
The default OS is ubuntu. The version is the latest stable version (e.g., 16.04
for now).
For -cuda version, the **cudnn** is included by default. Their versions are
also the latest stable version, i.e., cuda-8.0 and cudnn-5 for now.
We can implement this ticket in multiple steps. In particular, we separate the
images into multiple categories and create them one by one.
A:
singa:devel
singa:devel-cuda
singa:runtime
singa:runtime-cuda
B:
singa:devel-centos7-cuda
singa:devel-ubuntu14.04
singa:devel-ubuntu14.04-cuda7.5-cudnn4
...
Reference:
https://github.com/apache/incubator-singa/tree/master/tool/jenkins/docker
https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/docker/
> Create docker images
> --------------------
>
> Key: SINGA-276
> URL: https://issues.apache.org/jira/browse/SINGA-276
> Project: Singa
> Issue Type: Improvement
> Reporter: Moaz Reyad
>
> Docker can be used to improve the usability of SINGA by creating images which
> are ready to use.
> Following SINGA-274, the images can be
> singa: the core package for singa C++ on CPU
> singa-python: the pysinga, depends on singa pacakge on CPU
> singa-gpu : the gpu version of the singa C++, depends on singa and extends it
> with the gpu features
> singa-gpu-python: pysinga and gpu together. depends on singa-python and
> singa-gpu
> singa-dlaas: depends on singa-python, adds the DL as a service (Rafiki) on CPU
> singa-gpu-dlaas: depends on singa-gpu-python, adds the DL as a service
> (Rafiki) on GPU
> GPU packages can be split into cuda and opencl.
> Images should be uploaded into DockerHub
> The images could be generated automatically from GitHub, see
> (https://docs.docker.com/docker-hub/github/). Each image will have a Docker
> file and make script to build it automatically with every change in GitHub.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
