Proposal

Narrative

Faculty at the School of Architecture are already engaging and applying digital technologies with Artificial Intelligence and Virtual Reality to educate students, preparing them to create future novel achievements in architectural design. There are currently 10 Architecture courses that need access to these servers to support existing coursework related to developing complex simulations at varying scales, including: AI training, parametric design models, and augmented reality models that allow for real-time interactions between users and the models. In addition to the School of Architecture’s technology offerings, the School of Communication and Multimedia Studies offers 5 courses, capped at 18 students per course that use high-end computing for augmented and virtual reality coursework. With nearly 300 students from multiple programs using individual workstations, students are significantly limited by the technology currently available. Tasks involving deep learning training with large datasets for image models are time and computation expensive. Long processing times for large files force students to be grouped to account for computing limitations. During classes, faculty are unable to complete live demos, inhibiting student perception of industry-standard workflows. The OIT team in Jupiter is proposing to expand their server offerings to provide academic support for that will dramatically expand student access to high-end servers for coursework The OIT team in Jupiter has created “Onboarding”, a user-friendly interface for the servers, that will allow students to easily and remotely login to a computer, reserve a server node and get to work within minutes! The collaboration with Jupiter will allow the servers to float to assist other research programs at FAU and float back to help programs in Fort Lauderdale if additional resources are needed.

Facilities

The Office of Information Technology (OIT) at Florida Atlantic University (FAU) proposes expanding the research computing cluster located on the Jupiter campus. With dedicated compute nodes providing support to Fort Lauderdale academic services. Over the last year, the research computing cluster in Jupiter has provided 16.5 million CPU hours of computation, assisting 210 researchers and students, but requests for computation time are constant, especially in the areas of data science, machine learning, and artificial intelligence. Jupiter has pre-established clusters and the necessary resources, physical space, cooling, and infrastructure to support such a deployment. OIT has the ability to stand up the expansion within days of receipt of hardware and to bring additional services online as they are tested and completed. Having deployed the largest machine learning V100 GPU accelerated cluster at a University in Florida two years ago, OIT in Jupiter has the skills and expertise to hit the ground running and provide an immediate impact to researchers at FAU.

Hardware Requirements

The Research Computing team in OIT proposes the purchase of 3 units of an A100 based GPU accelerated compute cluster containing 46 individually assignable GPU cores. These machines would be made available to students and researchers in the School of Architecture and when not in use all of FAU as a whole. A total of 46 A100 GPU cores would allow School of Architecture researchers to compete with the best schools throughout the United States in a wide range of areas, including data science, 3D rendering, visualization and modeling, machine learning, and artificial intelligence. This project would enable your researcher groups both big and small to take advantage of cutting-edge technology for years to come.

Software Requirements

The OIT team in Jupiter has created “Onboarding”, a web-based, user-friendly interface for the servers, that will allow students to easily and remotely login to a computer, reserve a server node and get to work within minutes! The collaboration with Jupiter will allow the servers to float to assist other research programs at FAU and float back to help programs in Fort Lauderdale if additional resources are needed. Software that makes use of the servers are purchased though the School of Architecture, or other programs operating out of Fort Lauderdale.

Personnel Costs

None.

Other Costs

There are no other costs associated with this project. In this agreement between the Fort Lauderdale and Jupiter campuses, OIT is waiving all recurring costs to programs in Fort Lauderdale, such as hosting.

Timeline

March 2022 – approval of project April 2022 - Procure all equipment May - June 2022 - Contingent on equipment availability and shipping times. Install servers in Jupiter. July 2022 - Faculty training to use servers. August 2022 - Servers are fully operational and ready for instructional use beginning Fall 2022.

Sustainability

This project has been designed for long-term support and flexibility to support multiple programs at FAU.

Resource Matching

None.

Implementing Organization

This project will be implemented by the Office of Information Technology (OIT) at Florida Atlantic University's Jupiter campus.