Skip to content

Facilities, Equipment, and Other Resources: Research Computing at Oregon State University

Overview

The Research Computing Office (RCO) at Oregon State University provides shared high-performance computing (HPC), research data storage, training, and research software engineering (RSE) and consulting services to researchers across the university. The RCO's research infrastructure is built on "Wildwood," a campus research computing environment with a resilient service architecture, a large Slurm HPC cluster, and secure networks.

Wildwood is housed in multiple climate-controlled data centers across campus and is operated by the RCO. It supports a broad range of computational needs across disciplines and is continually updated. External collaborators can use it through a sponsoring OSU investigator.

Wildwood HPC Infrastructure

Resource Capacity
Data centers 3 separate buildings, climate controlled
Compute nodes 145 (32 with GPUs)
CPUs ~7,300 (~13,000 threads)
Aggregate memory ~86 TB
GPUs 122
Core systems Redundant management nodes, virtual machine infrastructure, control and utility services
Research data storage 20+ PB (see Research Data Storage)
Research backbone 100 Gbps to the campus research DMZ and Internet2
Data-handling network 40 Gbps for sequencing, backup, and data file servers
Compute node network 1/10 Gbps for all nodes, 40 Gbps for some

Scheduling and access

Slurm accounts and partitions control access. Fairshare, job-age, job-size, and partition-priority weighting balances the workload across groups. Users run batch and interactive jobs through SSH, VS Code remote tunnels, or JupyterHub, authenticating with OSU credentials. Dedicated file-transfer servers keep data movement off the login nodes.

Groups that purchase dedicated nodes receive priority access to them through lab and departmental partitions, while idle capacity remains available to the wider community through preemptible scheduling. GPU nodes [including NVIDIA V100 and T4 systems] are available to OSU researchers at no charge.

Software environment

Wildwood runs Rocky Linux 9 and provides more than 200 centrally maintained scientific applications through Lmod modules. Singularity/Apptainer containers and user-managed Pixi and Conda environments support reproducible research. Curated Nextflow workflows with cached container images support bioinformatics, and a Spack-built software stack serves earth, ocean, and atmospheric science workloads.

The RCO develops and maintains hpcman, a command-line tool for job submission, queue inspection, software setup, and lab and storage management.

Research Data Storage

Home directories

All OSU researchers with Wildwood access receive 100 GB of network storage at no charge, with nightly backups.

Tier 3 research data storage

Tier 3, mounted at /storage/t3, is the long-term home for research data on Wildwood. It is a multi-data-center Ceph system using erasure coding for resilience, and it serves active computing as well as long-term retention. Labs purchase capacity beyond the free allocation through RCO support.

Weekly Ceph snapshots protect the data, and an optional nightly backup of user-specified directories is available at the same rate as Tier 3 storage.

Backup and disaster recovery

Home directories receive daily incremental backups at no cost. Tier3 storage receives weekly snapshots saved for 3 weeks. Nightly backup of purchased space is available as a service.

Research and Academic Networking

The campus core network includes dual, redundant 100 Gbps paths to all science, research, and data center buildings, up to 10 Gbps to workstations, and wireless networking in common areas and classrooms at speeds up to 1 Gbps. Within Wildwood, file servers connect at up to 100 Gbps, control and utility services and some compute nodes at up to 40 Gbps, and the remaining compute nodes at 1 to 40 Gbps.

Training and Teaching: Advanced Cyberinfrastructure Teaching Facility (ACTF)

The RCO hosts a partitioned set of HPC resources within Wildwood dedicated to instruction in linux-based data science, along with enterprise-grade web hosting. Dedicated staff provide training on data management and curation, programming and software development, and analysis of next-generation sequencing data (RNA-Seq, GBS-Seq, and metagenomics). OSU instructors can use the ACTF for their linux-based courses.

Research Software Engineering and Consulting

RCO staff work directly with research teams to make computing more effective:

  • Consultations on workflow design, job optimization, parallelization, and efficient use of GPUs and storage
  • Software installation and containerization for reproducible environments
  • Pipeline development and migration, including Nextflow workflows and moving analyses from laptops or local servers to HPC
  • Proposal support, including resource estimates, data management plan language, and letters of support

Requests go through the RCO ticketing system.