Most research computing at UCR runs on the HPCC cluster. Cloud and national resources fill gaps: AI services, cloud-native tools, systems we do not have on campus, or more capacity than campus can offer. If you are not sure where to start, ask us. A short description of your work is enough.
Choose by what you are doing
| If you need to | Start with | Also consider |
|---|---|---|
| Run batch simulations, pipelines or MPI jobs | HPCC cluster | NSF ACCESS for larger runs |
| Train or run models on GPUs | HPCC cluster | NAIRR Pilot, NSF ACCESS, Nautilus |
| Use generative AI models in your research | Ursa Major (Tier 1 AI access) | Cloud and AI |
| Use exotic hardware the HPCC does not have (such as TPUs or Arm) | Ursa Major (Tier 1 exotic hardware, by consultation) | NSF ACCESS |
| Run thousands of small independent jobs | HPCC cluster | OSG / OSPool, with the OSDF for larger files |
| Do networking or distributed-systems research on real wide-area networks and programmable hardware | FABRIC testbed | NRP Nautilus |
| Run computer-systems experiments that need whole machines with root access (custom OS, bare-metal GPUs, FPGAs, Arm) | Chameleon testbed | FABRIC testbed |
| Run containers or JupyterHub at scale | NRP Nautilus | HPCC cluster (Singularity containers, Open OnDemand), Ursa Major (recharged) |
| Have a server or VM of your own (always-on web app, database, study server, Jupyter or RStudio server) | Jetstream2 (through NSF ACCESS) | Ursa Major (recharged), Cloud accounts |
| Have your own cloud account on a grant | Cloud accounts | Ursa Major |
| Work with regulated data | Secure research enclave | Security and data |
| Get help running a lab-owned cluster | RCSAS | HPCC cluster (lab-owned hardware in the campus cluster) |
Campus
HPCC cluster
The campus shared cluster for batch jobs, MPI, pipelines and GPU work, scheduled with Slurm. Lab subscription.
Ursa Major (Google Cloud)
AI model access, exotic hardware not available on campus, and archive storage. Other cloud work as recharged projects.
Cloud accounts (AWS, GCP, Azure)
Your own cloud account under the University of California agreements, billed to your funds through ITS.
Lab system administration (RCSAS)
Design and administration of lab-owned clusters and research servers, under an MOU.
National and shared
National programs award time by application rather than by recharge. They have their own policies on eligibility and data.
NSF ACCESS
National allocations on NSF-funded systems, by application. We can help you apply.
Jetstream2
NSF research cloud: Linux VMs you fully control, including NVIDIA A100 GPU VMs, through ACCESS.
NAIRR Pilot
The National AI Research Resource pilot: competitive allocations for AI research.
NRP Nautilus
A shared Kubernetes platform for research and education, run by the National Research Platform.
OSG / OSPool
Distributed high-throughput computing for many small, independent jobs, including GPU jobs.
Open Science Data Federation (OSDF)
National data origins and caches that deliver datasets to OSPool jobs, notebooks and laptops.
Open Storage Network (OSN)
NSF-funded S3 object storage, allocated through ACCESS, for sharing data across institutions and platforms.
FABRIC testbed
NSF testbed for networking and distributed-systems research: build your own multi-site networks of VMs, GPUs, FPGAs and programmable network hardware.
Chameleon testbed
NSF testbed for computer science: reserve whole bare-metal servers (GPUs, FPGAs, Arm) with root access, or VMs, for up to a week at a time.
A note on capacity
All of these are shared or allocated resources. Capacity, queue times, quotas and eligibility are set by each provider and change with demand. Choosing a service is not a reservation of capacity on it.