If you have any questions regarding ECI services or support, please contact us at help@engineering.ucsb.edu

ECI GPU Resources & Support

Whether you are teaching a course that requires parallel processing or setting up a high-performance system for your research lab, Engineering Computing Infrastructure (ECI) provides both managed hardware access and custom technical support.

Instructional GPU Cluster

ECI manages a dedicated instructional cluster funded by the College of Engineering to support coursework requiring parallel processing and heavy computation.

Cluster Resources: 42 compute nodes provided via Multi-Instance GPU (MIG) partitioning across six 80GB NVIDIA A100 GPUs.

Per-Node Allocation: Each node provides 10GB of GPU VRAM and 48GB of System RAM.

Workload Management: Jobs are submitted and managed using the Slurm workload manager.

Access Points: Students and TAs can access the Slurm environment directly from any ECI Linux computer lab.

How to Request Access for Your Class

To use the instructional cluster for your course, instructors must register their class with ECI prior to or at the start of the term.

Please submit your request with:

  •     Course name, number, and quarter.
  •     Complete roster of enrolled students.
  •     List of TAs and instructors requiring access.

Register Your Course for GPU Access 

Custom Research GPU Solutions

Because research requirements vary widely across departments, ECI offers bespoke infrastructure and administration services tailored to your lab's specific budget and computational needs.

While ECI does not directly fund research hardware, we partner with faculty and lab groups through every step of the infrastructure lifecycle:

Hardware Procurement & Specifying: Expert guidance on selecting and sourcing the right GPU hardware and node configurations for your research domain and budget.

System Deployment: Turnkey solutions ranging from bare-metal setup for direct faculty management to fully configured research clusters.

Queue & Cluster Management: Provisioning of custom environments, including single-node clusters or dedicated Slurm hosts—taking cluster administration off your hands so you can focus on your research.

Request a Research Consultation

Need help spec'ing a new system, configuring hardware, or setting up queue management for your lab?

Request a Research GPU Consultation