Webinar: JASMIN's new GPU cluster, ORCHID
We hosted a webinar on Tuesday 31st January 2023 at 10:30 about JASMIN’s new GPU cluster, ORCHID. The event provided information about ORCHID itself, as well as hearing from recent test users about their experiences with it. Users of other research GPU clusters also talked about their use cases and experience.
The webinar was open to all JASMIN users, but was of particular interest to those embarking on research involving machine learning techniques, or who have been seeking GPU resources for use in their work. Further details, including links to slides and recordings can be found below.
The event was successful with over 100 attendees and plenty of positive feedback about both the webinar content and the new GPU cluster itself.
Further webinars and workshops will be planned to allow discussion of issues and sharing recommendations that might benefit potential new users of the GPU - watch this space!
Agenda of the webinar
The agenda, slides and video recordings of the presentations are provided below. Some sections, such as the welcome and discussion, were not recorded.
- Welcome and opening - JASMIN facility director, Adrian Hines
- JASMIN’s new GPU cluster, ORCHID,Part 1: Introduction & overview, slides, presentation
- Experience from JASMIN GPU users:
- Training and running deep learning models on meteorological data using JASMIN -Jonathan Coney, University of Leeds, slides, presentation
-
IceNet: Sea ice forecasting training on Orchid, James Barnes, BAS, slides, presentation
-
A Flood Hazard Impact Model for India: Cylc workflows on JASMIN GPUs and CPUs, Steven Wells, CEH, slides, presentation
-
GPU training of CNNs for automatic industry detection in satellite imagery, Peter Manshausen, University of Oxford, slides, presentation
-
Optimising Python code to make use of GPUs – Nicola Martin, Plymouth Marine Laboratory, slides, presentation
-
Environmental enzymes - designing biochemistry to tackle environmental problems: a workflow with hierarchical configuration and direct submission on orchid", Simon Mathis, Cambridge University, slides, presentation
- Training and running deep learning models on meteorological data using JASMIN -Jonathan Coney, University of Leeds, slides, presentation
- Experience from other GPU services: The NEODAAS Artificial Intelligence Service – Dan Clewley, Plymouth Marine, slides, presentation
- JASMIN’s new GPU cluster, ORCHID - Part 2: What next? slides, presentation
- Wrap-up discussion (Transcript tba)
