Difference between revisions of "Draft:Software Installation Policies and Guidelines"

From UFRC
Jump to navigation Jump to search
m (Israel.herrera moved page Draft:Software Introduction to Draft:Software Installation and Policies Guidelines without leaving a redirect)
 
(21 intermediate revisions by 3 users not shown)
Line 1: Line 1:
General Discussion
+
==Generalized vs Specialized vs Domain Specific environments==
*User login server abuse—what to do? TODO: post a terminator whitelist to #apps and try
+
*Generalized environments like R and Python provide a wide range of functionality for data analysis, machine learning, and other scientific computing tasks. These environments have a large number of libraries and packages available for users to work with.
running it manually before letting it loose via a The login nodes might be big
+
*Specialized environments like PyTorch and TensorFlow are focused on specific machine learning tasks, such as deep learning, and provide specialized tools and libraries to support these tasks.
enough that we have a large gap between the resources cgroups_py tries to keep
+
*Domain-specific collections like GeoPython or DJPytt are environments that are tailored to specific domains, such as geospatial data analysis or data journalism.
available and our login policy. Checkout the 'arbiter' script again. Describe the
+
==Recommendations on using our prebuilt environments vs building your own==
functionality of the main R and python environments vs more specific environments like
+
Using prebuilt environments can be beneficial for users who want to get started quickly and don't need to make customizations to the environment. These environments are usually well-maintained.
pytorch and tensorflow or domain-specific collections like geopython or DJppytt)Q0
+
*Building your own environment can be beneficial if you need control over the environment or want to customize it for your specific needs. However, this can also be more time-consuming and requires more technical expertise.
*Published policies/guidelines regarding software installs? R looks like we don't have one.
+
===Prebuilt Environments for Reference===
would be useful to at least write out guidelines on the help site and set some limits and
+
There are many prebuilt environments available for popular libraries and frameworks, such as PyTorch. Users can refer to these environments for guidance on how to build their own environment or use prebuilt containers.
use expectations.  
+
*For example, NVIDIA provides a prebuilt PyTorch container that users can use as a starting point for their own environment.
  
TODO: write a help page on software installation guidelines and limits.
+
==Availabile Linux Software installed in ResVault==
 
+
* ResVault VMs have access to the Linux application installed on HiPerGator in /apps.
 
+
* To install custom applications into a ResVault VM, please contact support via [https://my.it.ufl.edu/ Cherwell]
Things we want to see on that page
+
__NOTOC__
* Recommendations on using our prebuilt e.nys vs building their own in the context
 
of speed of customization vs. maintaining functional environments. Focus on R
 
and python at first.
 
* Curated list Of prebuilt environments for users to reference in the context Of the
 
above recommendations.
 
* Published environment.yml files for the popular environments like pytorch, which
 
users could refer to when starting to build their or use prebuilt containers
 
e.g. pytorch container from NVIDIA.
 
* Availability of the linux software installed in lapps in ResVault.
 

Latest revision as of 18:47, 20 February 2023

Generalized vs Specialized vs Domain Specific environments

  • Generalized environments like R and Python provide a wide range of functionality for data analysis, machine learning, and other scientific computing tasks. These environments have a large number of libraries and packages available for users to work with.
  • Specialized environments like PyTorch and TensorFlow are focused on specific machine learning tasks, such as deep learning, and provide specialized tools and libraries to support these tasks.
  • Domain-specific collections like GeoPython or DJPytt are environments that are tailored to specific domains, such as geospatial data analysis or data journalism.

Recommendations on using our prebuilt environments vs building your own

Using prebuilt environments can be beneficial for users who want to get started quickly and don't need to make customizations to the environment. These environments are usually well-maintained.

  • Building your own environment can be beneficial if you need control over the environment or want to customize it for your specific needs. However, this can also be more time-consuming and requires more technical expertise.

Prebuilt Environments for Reference

There are many prebuilt environments available for popular libraries and frameworks, such as PyTorch. Users can refer to these environments for guidance on how to build their own environment or use prebuilt containers.

  • For example, NVIDIA provides a prebuilt PyTorch container that users can use as a starting point for their own environment.

Availabile Linux Software installed in ResVault

  • ResVault VMs have access to the Linux application installed on HiPerGator in /apps.
  • To install custom applications into a ResVault VM, please contact support via Cherwell