Is it right for you?
A good fit
- Batch jobs you can queue and walk away from
- Many-core or multi-node MPI work
- GPU training and inference run as batch jobs
- Pipelines with many independent tasks
Not a fit
- Data rated P3 or P4 (see the secure enclave)
- Always-on web services or databases
- Work that needs a dedicated machine at a fixed time
- Long-term storage of finished data
What you get
The High-Performance Computing Center (HPCC) runs UCR's shared research cluster. Lab members share access to its CPU and GPU partitions, a large library of installed software, parallel storage and HPCC workshops. The HPCC currently describes the cluster as over 16,000 CPU cores and NVIDIA GPU nodes (HPCC Access page and Recharging Rates 2026/2027, as of Oct 2026), with over 1,200 software tools and 50 databases (HPCC Recharging Rates 2026/2027, as of Oct 2026).
Hardware, partitions and software change over time. The HPCC's hardware pages are the authoritative and current description.
Costs
Access is through an annual lab registration paid from a UCR funding source: $1,000 per lab per year (HPCC Recharging Rates 2026/2027, as of Oct 2026). Additional storage can be rented ($1,000 per 10 TB per year, or $25 per 100 GB per year) or bought as lab-owned disk with an annual maintenance fee. A labor rate applies to work beyond the included consultation.
Rates for external collaborators differ. The HPCC Recharging Rates document is the authority on every figure here, and the rate sheet in force when you are billed is the one that applies. See Costs for how recharge works across services.
Limits and fair use
The cluster is shared. The HPCC sets per-user and per-lab CPU quotas: currently 384 cores per user, 768 cores per lab (HPCC Recharging Rates 2026/2027, as of Oct 2026). Jobs over a quota are accepted but wait in the queue until resources within the quota become available. Queue priorities and maximum run times also apply and may be adjusted by the HPCC as demand changes. Start times for queued jobs depend on demand and are not predictable.
How to get access
Account requests go to the HPCC by email from the PI, or with the PI copied, as described on the HPCC Access page. A lab that is not yet registered provides a funding source (COA) for the annual registration. Once the account exists, follow the HPCC login instructions.
Connect and run
You sign in over SSH and submit work with Slurm. The HPCC manuals cover logging in, transferring data, the module system, and writing job scripts. Moving work from a cloud VM to the cluster is covered in KB022: Migrating workloads from Google Cloud to HPCC.
Get help
For cluster questions (accounts, jobs, software installs), contact the HPCC directly through the channels on their site. For help choosing between the HPCC and other options, or planning a project, contact Research Computing.