If your research has outgrown your laptop or lab workstation, you do not need to buy a server. University of Windsor faculty members, and the students and postdocs they sponsor, can use Canada's national research computing systems at no cost through SHARCNET and the Digital Research Alliance of Canada. Paul Preney, UWindsor's SHARCNET consultant, can meet with you on campus or online to help you choose a system, set up an account and get your code running.
Ready to start? Get an account, find the right resource or email Paul Preney to book a consultation.
On this page
- Find the right resource
- Get an account
- National systems you can use
- Who provides what
- Training and events
- Researcher stories and common questions
- Get help
Nibi at a glance
Nibi is the national cluster hosted and run by SHARCNET at the University of Waterloo.
136,000
CPU cores
312
GPUs for AI and accelerated code (288 NVIDIA and ~24 AMD GPUs)
500 GB to 6 TB
memory on Nibi's nodes
No cost
for eligible faculty and the users they sponsor
Find the right resource
Choose what you need to do. Each option points you to where to start.
Run large simulations or many jobs
Start on Nibi, which suits most simulation and data analysis work. Submit many independent jobs at once with Slurm job arrays.
Train AI or machine learning models
Use the GPU nodes on Nibi, Fir or Rorqual. PyTorch, TensorFlow and JAX are available as modules or can be set up in a virtual environment. Large or long training campaigns may need a Resource Allocation Competition application.
Host a web service or virtual machine
Request an Alliance Cloud project by submitting a ticket to Alliance technical support. Paul Preney can help you size your request.
Store and share large datasets
Every group gets shared project space on each cluster, and nearline storage holds large datasets you use less often.
Learn the basics first
Take a self-paced SHARCNET course, watch past seminars, or book a meeting with Paul Preney to get your first job running.
Get an account
Accounts are managed through the Alliance's CCDB (Compute Canada Database). Faculty members register first, then sponsor their research group.
- Faculty member registers. Principal investigators create an account at ccdb.alliancecan.ca using their UWindsor email.
- Students and postdocs join. Group members register and list their supervisor as sponsor. The supervisor approves each request.
- Log in and run jobs. Connect to a cluster with SSH or through a web browser, then submit work through the Slurm scheduler.
- Ask for more if you need it. Default allocations suit most groups. Larger projects apply to the annual Resource Allocation Competition.
Read the full instructions on how to apply for a CCDB account in the Alliance documentation.
National systems you can use
Once you have an account, you can log in to any of these clusters. Cloud access is requested separately through a support ticket. Each system name links to its full hardware details.
Quick guide by type of work
- CPU work: Nibi, Fir, Rorqual, Narval
- GPU and AI work: Nibi, Fir, Rorqual, Trillium, Narval
- Large parallel jobs across many nodes: Trillium
- Cloud services and virtual machines: Alliance Cloud
Nibi
Location: University of Waterloo, run by SHARCNET. Ontario’s home cluster.
General-purpose cluster with about 136,000 cores, 312 GPUs (288 NVIDIA and ~24 AMD GPUs) and nodes with 500 GB to 6 TB of memory. The natural first stop for Ontario researchers.
Best for: CPU, GPU and large-memory work
Fir
Location: Simon Fraser University, British Columbia
Large general-purpose cluster for CPU and GPU work, including long-running and high-throughput jobs.
Best for: CPU and GPU work
Rorqual
Location: Montréal, run by Calcul Québec
General-purpose cluster with strong GPU capacity for machine learning and accelerated codes.
Best for: CPU and GPU work
Trillium
Location: University of Toronto, run by SciNet
Built for large parallel jobs that span many nodes at once, such as climate, fluid dynamics and physics codes.
Best for: large parallel and GPU work
Narval
Location: Montréal, run by Calcul Québec
General-purpose cluster for CPU and GPU workloads.
Best for: CPU and GPU work
Alliance Cloud
Location: National cloud sites, including Arbutus at the University of Victoria
Virtual machines for web portals, databases, research platforms and services that need to stay online. Request access with a support ticket.
Best for: cloud services
Who provides what
Four organizations work together so a UWindsor researcher can go from a first question to a national-scale allocation.
Campus: University of Windsor
Consulting from Paul Preney, on campus or online. Help with accounts, software and getting started.
Regional: SHARCNET
A consortium of about 19 Ontario institutions. Runs Nibi, training and technical support.
Provincial: Compute Ontario
Coordinates advanced research computing across Ontario.
National: Digital Research Alliance of Canada
Funds and operates the national systems, storage and cloud.
Training and events
Free courses for faculty, staff and students, from your first Linux command to GPU programming.
- SHARCNET online training: self-paced courses and recorded workshops on the SHARCNET training website.
- General Interest Seminars: regular webinars on tools, languages and new systems. Recordings are on SHARCNET's YouTube channel.
- Summer school: annual intensive workshops on HPC, AI and data science, run by Compute Ontario with SHARCNET and the other Ontario consortia. See the Compute Ontario training website for details.
- Annual Introduction to SHARCNET talks: slides from Paul Preney's yearly UWindsor talk, and SHARCNET's other introductory presentations, are online. On the SHARCNET training website, select Getting Help, then the current year's Introduction to SHARCNET Presentations, and open the UWindsor section.
Researcher stories and common questions
Dr. John Trant's team in the Department of Chemistry and Biochemistry uses computational design to guide the creation of new drugs and materials, and is one of UWindsor's significant users of national research computing resources. The team built an open-access tool that screens every possible short peptide, a library of 3.2 million combinations, against a target protein. In one test, they screened every combination against a key protein of the virus that causes COVID-19 in a single afternoon.
Learn more on Dr. Trant's UWindsor faculty page or read the DailyNews story Computational tool to simplify process of drug discovery.
Dr. James Gauld and his team in the Department of Chemistry and Biochemistry use SHARCNET's high performance computing to study how biological catalysts work. Their computational chemistry models look at enzymes such as nitric oxide synthases, which produce nitric oxide, a molecule involved in brain development and blood pressure regulation. The team also studies ribozymes, RNA molecules that can act as catalysts.
Learn more on Dr. Gauld's UWindsor faculty page.
Any UWindsor faculty member who leads research can register. Graduate students, postdocs, research staff and undergraduate research assistants can register when a faculty member sponsors them.
No. Eligible Canadian academic researchers use the national systems free of charge. Every group receives a default allocation, and groups that need more compute or storage can apply to the annual Resource Allocation Competition.
Far too many to list here. Hundreds of scientific packages, compilers and libraries are installed and loaded with the module command. Most are open source, and some are proprietary or commercial. See the full list of available software. You can also install your own software or use containers.
Before moving any sensitive or confidential data, submit a ticket describing your project and data to support@tech.alliancecan.ca.
Yes. You can start with our introductory courses on the SHARCNET training website and our technical documentation wiki, and our systems can be accessed using a web browser. If you would like, arrangements can be made for Paul Preney to meet with you and/or your team to help you get things running.
Get help
Start with campus support. For technical problems on a national system, the Alliance support desk can respond directly.
Paul Preney
High Performance Computing Technical Consultant / Research Associate, University of Windsor and SHARCNET. Meets on campus or online.
Book a consultation by email: preney@uwindsor.ca
Alliance technical support
Account issues, job errors and system questions on national clusters.
Email: support@tech.alliancecan.ca
Learn more on the SHARCNET website, the Compute Ontario website, the Digital Research Alliance of Canada website and in the Alliance technical documentation.