Is it right for you?
A good fit
- Generative AI model access for research and programming
- Work that needs exotic hardware the HPCC does not have, such as TPUs or Arm processors
- Long-term archive of data you must keep but rarely read
- Cloud-native work (VMs, containers, databases, analytics), as a recharged project
Not a fit
- Batch or GPU computing that runs well on the HPCC
- Regulated data outside an approved environment (see the secure enclave)
- Individual accounts not tied to a lab or PI project
- Projects expecting Research Computing to fund recharged usage
What it is
Ursa Major is UCR's research program on Google Cloud, run by Research Computing with ITS. It focuses campus support on what campus systems cannot easily provide: access to AI models for research, exotic hardware the HPCC does not have, and long-term archive. Other cloud work is available as recharged projects. It complements, rather than replaces, the campus cluster.
How allocation works
Requests are reviewed against current campus resources, funding and research priorities. Resources fall into tiers:
- Tier 1, campus-supported. Three things, with no recharge to the lab under current terms, within limits and subject to eligibility:
- AI model access for research and programming, through a service Research Computing manages, with a per-lab allowance;
- exotic hardware not available on the HPCC, such as TPUs or Arm processors, through a Research Computing HPC cluster in Google Cloud, within limits set per project, starting with a consultation;
- archive storage in Google Cloud archive classes.
- Tier 2, recharge. All other cloud work, including VMs, GKE, Cloud SQL, BigQuery, Standard storage, GPUs and marketplace models, billed to a lab funding source through ITS.
- Tier 3, dedicated agreements. Very large or multi-year projects may need their own contract with the provider, with ITS oversight.
KB005: Ursa Major service tiers describes the tiers in more detail.
$135 one time and $407 per year administrative fees can apply to recharged cloud accounts. These are set out in the MOU for the account.
Costs and campus support
Tier 1 is funded from campus sources and depends on continued funding. Research Computing does not commit to how long any service will remain without recharge. Lab-funded (recharged) use is billed at the rates in your MOU.
Projects set up before October 2026 may still be arranged under the earlier tiers. See the archived description.
How to get started
Contact Research Computing with a short description of the project, the data involved (and its protection level), which Tier 1 area you need (if any), and a funding source if recharged work is likely. Student requests are approved by the student's PI and placed in the PI's project.
Related
- AI at UCR for campus AI tools and where their terms are published.
- Cloud archive for long-term data retention in Google Cloud.
- KB007: Ursa Major recharge workflow for setting up a recharged project.