Authors: Jeff Goodwin, Nick Karker
Description
The AI Supercomputer SLA outlines the terms and conditions for accessing and utilizing the NVIDIA DGX Super Pod, referred to as the "AI Supercomputer." This infrastructure provides high-performance AI computing capabilities for conducting university research, teaching and learning. The service level agreement is aimed at UAlbany researchers who require advanced computational resources.
Available To
Please see the section here to see the current availability to different user groups
Features
...
Users may submit up to 4 jobs at a time. While up to 4 jobs can be submitted, only 1 of those jobs will accrue priority and run at a time.
A single job can request up to 4 GPUs
A job will run up to 8 hours
Resource Availability
Non-priority access
...
provides a maximum of 4 GPU’s and 1TB of high-performance all-flash storage
...
. In unique situations, reservations for additional resources may be granted as resource availability allows.
Service Availability
The AI Plus Supercomputing Cluster is technically available 24 hours a day, 7 days a week, and 365 days (12 months) a year (24x7x365). While free access is subject to availability and is not guaranteed, the University at Albany is committed to providing reliable access to the NVIDIA cloud Infrastructure to support all research endeavors. We aim to manage resources effectively to maximize availability for both priority and non-priority users
...
Resource Allocation: Access limited to a maximum of 4 GPU’s and 1TB of high-performance all-flash storage.
Service Availability
The AI Plus Supercomputing Cluster is technically available 24 hours a day, 7 days a week, and 365 days (12 months) a year (24x7x365).
Exceptions
Scheduled Maintenance: Scheduled maintenance windows will be communicated in advance. During these periods, the infrastructure may be temporarily unavailable.
Emergency Downtime: In an emergency, the infrastructure may be unavailable. Efforts will be made to restore services as quickly as possible.
...