Customer Story • Higher Education

UC Santa Barbara AgilityFlex Cluster

A 300-teraFLOP shared research platform for a growing computational campus resource

UC Santa Barbara needed a modern shared cluster that could support traditional simulation along with data-intensive research, machine learning, and emerging computational work across a broad set of disciplines.

The “Pod” system replaced an older 12-teraFLOP cluster with substantially more compute capability and modern GPU resources.

Request a Consultation

300

teraFLOPs of compute capability

$1.1M

Reported project value

27+

Departments and research units using shared clusters

12 → 300

teraFLOP step-up from the prior cluster

Research Requirements

One Cluster, Many Disciplines

A shared university platform must serve researchers whose codes, data sets, and computational patterns can be very different.

Traditional HPC

Engineering, chemistry, physics, materials, and simulation workloads continue to demand large-scale parallel compute.

Machine Learning & AI

Modern GPU resources expanded support for research groups adopting accelerated computing and data-driven methods.

Data-Intensive Science

Biology, earth science, psychology, and data science increasingly require compute beyond conventional workstations.

The Platform

“Pod” — A Shared Campus Research Resource

The 300-teraFLOP cluster is hosted by UCSB’s Center for Scientific Computing and supports scientific and engineering research across the university.

It replaced the older “Knot” cluster, which had supported hundreds of research publications and a wide range of disciplines. The new platform increased available compute capacity while adding modern accelerator resources for new classes of workloads.