High performance computing
From Robin
Background
Many of ROBIN's activities require computing power. Whether it is heavy simulations or complicated optimization, the need for computing power has only increased. In recent years, this need has been reinforced as the popularization of deep learning on GPUs.
Overview
Below we list the resources available at ROBIN and UiO:
- ROBIN local GPU-equipped workstations
- We have a selection of different local GPU-equipped workstations at ROBIN.
- The primary use-case for these machines is when you cannot use AI hub or Sigma2. For example if your work is on a physical setup/robot and/or you require realtime sensing. Alternativly, if you require specific software that is not available on central services.
- Talk to your supervisor or send an e-mail to robin-engineer at ifi.uio.no to get access.
- USIT AI hub
- Consists of 4 powerful computers with a lot of GPU computing power.
- Students will be granted access at the request of the supervisor.
- Sigma2 - National HPC resources including the Saga supercomputer
- Consists of a powerful cluster with CPU and some GPU resources.
- Makes use of job scheduling.
- Students will be granted access at the request of the supervisor.
- NREC
- This is a "Infrastucture as a service" and uses OpenStack to manage instances.
- Consists of CPU hardware only.
- A great service if you require root access to an instance.
- ROBIN has access to a 2x64 core machine with 512 GB of RAM. Still a WIP project.
