[SIG] Requests for SIG Proposal: SIA-AI #7

New Issue

brian · 2023-03-18T01:21:04Z

brian commented

2023-03-18 01:21:04 +00:00

On behalf of Balaji Balakrishnan:

Initial Project

SIG/AI

The idea is to promote usage of Rocky Linux among the AI, ML and Data Science community.

Promoting Interfaces

1. Social Media

Using Twitter, Facebook, Youtube, Medium.com – to publish short articles and videos frequently which emphasis on robust, secure, stable and performance of rocky linux.

2. Blog Articles

A brand new blog.rockylinux.com could be a good start to showcase rockylinux as an equivalent alternatives to other OS in usage now. We can publish Rocky Linux version of top rated blog articles which showcases the easy configuration and working proof using other OS.

3. Online Meetups and Event Sponsoring

We can build regional virtual communities in countries of high downloads. Give out digital swags to share in their profile for the meetups.

4. Community partners

we can contact other opensource projects and talk to them about publishing articles in their blog and update their documentation with installation steps for rocky linux. We can help them implement, write and ensure the documentation is authentic and workable, then publish. Tensorflow, Pytorch, and other ML tools, including W&B.ai, clear.ml. Neptune.ai are few to being with.

Engineering Interface

1. NVIDIA has already acknowledged us with the release of CUDA repo officially for Rocky Linux. Since we are bug to bug compatible with RHEL we can put some instructor videos on setting up clustering with rocky for multi-node, multi-gpu training/testing. Growing need for multi-node training keeps us ahead of other OS like ubuntu.

2. We already have cloud partners with us to prove the robustness, we need to target the laptop users with OS like Linux Mint initially, since they look for less maintenace robust OS. We can be a good alternative for enterprise users who needs security.

3. we have the advantage of having compatibility with many opensource Cluster Managers and job Schedulers like SLURM, Open PBS etc.... If we can shed lights on these areas tp setting up clusters at a research lab/group level. We can definitely add considerable number of users.

4. Maintainability is an aspect that needs to be considered when it comes to multi-user machines/clusters. We need to focus on giving users an ecosystem where they can learn to easily maintain a crowd of individual researchers/ research groups.

5. promoting the use of containerization using Nvidia NGC, Docker etc...

6. Getting the Opensource apps like W23 Opensource Awards from Y Combinator to publish the installation, usage and performance reports of their apps on Rocky Linux.

7. Creating the comparison reports for security, maintainability, backup, recovery would be a strength for rocky

8. Our technical efforts should also focus on proving that the switching to rocky is no lesser in terms of functionality, availability of apps in every day life (Libre Office, Online Meeting apps like teams, zoom, webex etc.., multimedia, customaization, look and feel)

9. we may have to test and publish articles on 3rd party names which talks about battery life, how to optimize rocky for development, every day use, students, teachers, researchers etc.... every user base needs to be brought into light. ( we may have to try latest compatible kernels, themes, desktop editions). I think our SIG will come handy.

Once we gain the network of opensource partners, we can keep promotions aside and focus on AI engineering of rocky with a large user base I have laid this plan for next 6 months. This may require some time effort from all of us. In my view this is possible and i request you to put your inputs in to it. Also, this might require some items in terms of a designated role in SIG/AI, email address and access to our social media etc..

Initial Asks

Mattermost channel - https://chat.rockylinux.org/rocky-linux/channels/sig-ai
Forums section - https://forums.rockylinux.org/c/sig/ai
Package repository - PENDING (see below)
SIG webpage(s) with blog and documentation section (needs discussion on best way to arrange this)
Email address for outreach efforts - sig-ai@rockylinux.org
Groups in Rocky Account Services - DONE
RESF Git Service Organization - DONE
Mail List - https://lists.resf.org/mailman3/lists/rocky-sig-ai.lists.resf.org/
Bridge Mattermost Channel to IRC and Matrix - PENDING (see below)
Rocky Linux GitLab groups - DONE

Checklist

I have read the Special Interest Group Guide
I have put in the initial proposal at the appropriate venue (such as mattermost)
I acknowledge that I may be sponsoring this Special Interest Group

On behalf of Balaji Balakrishnan: ## Initial Project > **SIG/AI** > > The idea is to promote usage of Rocky Linux among the AI, ML and Data Science community. > > **Promoting Interfaces** > > 1\. Social Media > > Using Twitter, Facebook, Youtube, Medium.com – to publish short articles and videos frequently which emphasis on robust, secure, stable and performance of rocky linux. > > 2\. Blog Articles > > A brand new blog.rockylinux.com could be a good start to showcase rockylinux as an equivalent alternatives to other OS in usage now. We can publish Rocky Linux version of top rated blog articles which showcases the easy configuration and working proof using other OS. > > 3\. Online Meetups and Event Sponsoring > > We can build regional virtual communities in countries of high downloads. Give out digital swags to share in their profile for the meetups. > > 4\. Community partners > > we can contact other opensource projects and talk to them about publishing articles in their blog and update their documentation with installation steps for rocky linux. We can help them implement, write and ensure the documentation is authentic and workable, then publish. Tensorflow, Pytorch, and other ML tools, including W&B.ai, clear.ml. Neptune.ai are few to being with. > > **Engineering Interface** > > 1\. NVIDIA has already acknowledged us with the release of CUDA repo officially for Rocky Linux. Since we are bug to bug compatible with RHEL we can put some instructor videos on setting up clustering with rocky for multi-node, multi-gpu training/testing. Growing need for multi-node training keeps us ahead of other OS like ubuntu. > > 2\. We already have cloud partners with us to prove the robustness, we need to target the laptop users with OS like Linux Mint initially, since they look for less maintenace robust OS. We can be a good alternative for enterprise users who needs security. > > 3\. we have the advantage of having compatibility with many opensource Cluster Managers and job Schedulers like SLURM, Open PBS etc.... If we can shed lights on these areas tp setting up clusters at a research lab/group level. We can definitely add considerable number of users. > > 4\. Maintainability is an aspect that needs to be considered when it comes to multi-user machines/clusters. We need to focus on giving users an ecosystem where they can learn to easily maintain a crowd of individual researchers/ research groups. > > 5\. promoting the use of containerization using Nvidia NGC, Docker etc... > > 6\. Getting the Opensource apps like [W23 Opensource Awards from Y Combinator](https://pbs.twimg.com/media/FprCQHraQAEIn2X?format=jpg&name=medium) to publish the installation, usage and performance reports of their apps on Rocky Linux. > > 7\. Creating the comparison reports for security, maintainability, backup, recovery would be a strength for rocky > > 8\. Our technical efforts should also focus on proving that the switching to rocky is no lesser in terms of functionality, availability of apps in every day life (Libre Office, Online Meeting apps like teams, zoom, webex etc.., multimedia, customaization, look and feel) > > 9\. we may have to test and publish articles on 3rd party names which talks about battery life, how to optimize rocky for development, every day use, students, teachers, researchers etc.... every user base needs to be brought into light. ( we may have to try latest compatible kernels, themes, desktop editions). I think our SIG will come handy. > > Once we gain the network of opensource partners, we can keep promotions aside and focus on AI engineering of rocky with a large user base I have laid this plan for next 6 months. This may require some time effort from all of us. In my view this is possible and i request you to put your inputs in to it. Also, this might require some items in terms of a designated role in SIG/AI, email address and access to our social media etc.. ## Initial Asks - [x] Mattermost channel - https://chat.rockylinux.org/rocky-linux/channels/sig-ai - [x] Forums section - https://forums.rockylinux.org/c/sig/ai - [x] Package repository - PENDING (see below) - [x] SIG webpage(s) with blog and documentation section (needs discussion on best way to arrange this) - [x] Email address for outreach efforts - sig-ai@rockylinux.org - [x] Groups in Rocky Account Services - DONE - [x] RESF Git Service Organization - DONE - [x] Mail List - https://lists.resf.org/mailman3/lists/rocky-sig-ai.lists.resf.org/ - [x] Bridge Mattermost Channel to IRC and Matrix - PENDING (see below) - [x] Rocky Linux GitLab groups - DONE ## Checklist - [x] I have read the [Special Interest Group Guide](https://wiki.rockylinux.org/special_interest_groups/sig_guide/) - [x] I have put in the initial proposal at the appropriate venue (such as mattermost) - [x] I acknowledge that I may be sponsoring this Special Interest Group

brian added the

SIG_Proposal

label 2023-03-18 01:21:04 +00:00

brian referenced this issue from infrastructure/meta

2023-03-18 01:23:00 +00:00

AI special interest group initial requirements #14

neil self-assigned this 2023-03-23 19:25:18 +00:00

neil commented

2023-03-28 18:52:02 +00:00

Package repo - will be taken care of during sync. Needs to be discussed and planned with releng.
Webpage - we can do this via a wiki repo in the Gitea (git.resf.org) space for the SIG. Once that is setup, we can deploy it.

Pending IRC setup and bridging

* Package repo - will be taken care of during sync. Needs to be discussed and planned with releng. * Webpage - we can do this via a `wiki` repo in the Gitea (git.resf.org) space for the SIG. Once that is setup, we can deploy it. Pending IRC setup and bridging

neil commented

2023-04-05 00:18:32 +00:00

wiki is setup https://sig-ai.rocky.page/