GPU labs that are ready when the cohort arrives.
Managed environments on Podstack Cloud: the same image for every student or researcher, fractional GPUs so a class shares a few cards, billed per minute so idle nights cost nothing. Or licence Podstack OS and host the labs yourself.
Environments eat the semester.
- Teams spend weeks building and babysitting GPU environments instead of doing science or teaching.
- Giving many students reproducible, governed access is painful: versions drift and every user hits a different wall.
- On-premises clusters are capital-heavy and sit idle between projects and semesters.
- Who can use what, for how long and on whose budget is rarely clear until the bill arrives.
One image, many users, billed by the minute.
Pre-configured environments
QuickPods templates with PyTorch, Jupyter and CUDA, plus course or project images you supply. Users open a browser and start working.
Reproducible and governed
Every user gets the same known-good image, with roles, quotas and budgets per cohort or lab.
Fractional GPUs, shared
PodVirt lets many users share a few cards, so a class of dozens runs without a GPU per head.
Labs cost only when used
Per-minute billing means an environment costs while a session runs and nothing overnight, on weekends or between terms.
Self-host if you need to
Run it managed on Podstack Cloud, or licence Podstack OS and host the labs inside your institution for residency and control.
What a lab includes.
| Spec | Detail |
|---|---|
| Environments | PyTorch, Jupyter, CUDA and custom templates |
| GPU models | The NVIDIA families Podstack Cloud offers, shared fractionally |
| Sharing | PodVirt fractional GPUs for many concurrent users |
| Access | Multi-user with roles, quotas and budgets |
| Billing | Per minute, only while a session runs |
| Audience | Universities, research groups, enterprise R&D |
| Deployment | Managed on Podstack Cloud or self-hosted on Podstack OS |
Questions, answered.
What is a High Performance Lab?
A managed GPU environment your class or team uses out of the box: users log in to a pre-configured image on shared fractional GPUs instead of building and maintaining machines.
How does it work for a whole class or research group?
Many users share pooled GPUs through PodVirt, each getting the same reproducible environment, while admins set roles, quotas and budgets for the cohort.
Do we pay for idle time?
No, labs are billed per minute and only while a session is running, so environments left idle overnight or between semesters do not accrue GPU charges.
Can we host the labs ourselves?
Yes, run them managed on Podstack Cloud or licence Podstack OS and host them inside your institution when residency or full control is required.
Set up a lab for the next term.
Tell us the cohort size and the courses; we size the templates and the budget with you.